Plumbing trouble repair cost estimation program

Abstract

To estimate a repair cost for trouble in wet areas.SOLUTION: A wet area trouble repair cost estimation program for estimating a repair cost for trouble in wet areas causes a computer to execute: an information acquisition step of acquiring trouble type information on the type of trouble in wet areas; and an estimation step of using a learned model in which three or more stages of the degree of association between the repair cost and trouble type information for reference on the type of trouble in wet areas acquired in advance are defined, and which has the trouble type information for reference as input and the repair cost as output, and based on the trouble type information for reference according to the trouble type information acquired in the information acquisition step, estimating the repair cost, while putting priority on one with the higher degree of association.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a plumbing trouble repair cost estimation program. [Background technology]

[0002] Examples of plumbing problems in architectural structures such as buildings, condominiums, and detached houses include clogged toilets, leaks, and malfunctions, clogged kitchens and bathrooms, clogged bath drains, leaks and malfunctions in showers and faucets, etc. As such plumbing problems in architectural structures are often unable to be resolved by the residents themselves, they often call in specialist companies to carry out the work.

[0003] However, in recent years, the labor shortage among plumbers has become more serious, resulting in an increasing number of cases where residents contact plumbers without receiving a response. In particular, in the case of a clogged toilet, the toilet cannot be used until the clog is unclogged, and delays in the plumber's arrival can cause significant disruption to daily life. The same applies to clogs and leaks in kitchens and bathrooms. Such plumbing problems often require prompt response. Furthermore, since it is often difficult to immediately determine the repair costs, repair time, and repair methods, residents may hesitate to request plumbing repairs. As a result, continuing to use plumbing despite damage or malfunction can result in continued inconvenience for the residents. Similarly, neglecting to perform repairs or maintenance in advance of potential malfunctions or breakdowns can lead to similar inconveniences. To prevent such situations, there has long been a need for technology that can instantly determine the repair costs, repair time, and repair methods for plumbing damage or malfunctions. Summary of the Invention [Problem to be solved by the invention]

[0004] Until now, no technology has been proposed that can instantly determine the repair costs, repair time, and repair methods for such buildings.

[0005] The present invention was devised in consideration of the above-mentioned problems, and its purpose is to provide a plumbing trouble repair cost estimation program that can accurately and automatically determine the repair costs, repair period, and repair methods for plumbing troubles without relying on human labor. [Means for solving the problem]

[0006] The water pipe deterioration degree determination program according to the present invention is a water pipe trouble repair cost estimation program for estimating repair costs for water pipe troubles, which includes trouble type information relating to the type of water pipe trouble and The above mentioned plumbing The method is characterized in that the computer is caused to execute an information acquisition step of acquiring contract information relating to the contract contents of the repair support contract for the trouble, and an estimation step of estimating repair costs using a trained model in which a combination of previously acquired reference trouble type information relating to the type of plumbing trouble and reference contract information relating to the contract contents of the repair support contract for the plumbing trouble is defined, the input of which is the reference trouble type information and the reference contract information, and the output is the repair costs, giving priority to the reference trouble type information corresponding to the trouble type information acquired in the information acquisition step and the reference contract information corresponding to the contract information. [Effects of the Invention]

[0007] Even without special skills or experience, it is possible to instantly predict the repair costs, repair period, and repair methods for plumbing problems within building structures without relying on human labor. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a block diagram showing the overall configuration of a system to which the present invention is applied. [Figure 2]FIG. 2 is a diagram illustrating a specific configuration example of a search device. [Figure 3] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 4] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 5] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 6] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 7] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 8] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 9] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 10] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 11] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 12] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 13] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 14] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 15] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 16] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 17] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 18] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 19] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 20] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 21] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 22] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 23] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 24] FIG. 10 is a diagram for explaining the operation of the present invention. [Figure 25] FIG. 10 is a diagram for explaining the operation of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0009] A plumbing trouble occurrence frequency prediction program and a plumbing trouble repair cost estimation program to which the present invention is applied will be described in detail below with reference to the drawings.

[0010] First embodiment 1 is a block diagram showing the overall configuration of a plumbing trouble occurrence frequency system 1 in which a plumbing trouble occurrence frequency prediction program according to the present invention is implemented. The plumbing trouble occurrence frequency system 1 includes an information acquisition unit 9, a search device 2 connected to the information acquisition unit 9, and a database 3 connected to the search device 2.

[0011] The information acquisition unit 9 is a device through which users of the system input various commands and information. Specifically, the information acquisition unit 9 includes a keyboard, buttons, a touch panel, a mouse, switches, etc. The information acquisition unit 9 is not limited to a device for inputting text information, but may also include a device capable of detecting voice and converting it into text information, such as a microphone. The information acquisition unit 9 may also be configured as an imaging device capable of capturing images, such as a camera. The information acquisition unit 9 may also be configured as a scanner capable of recognizing character strings from paper documents. The information acquisition unit 9 may also be integrated with the search device 2, which will be described later. The information acquisition unit 9 outputs the detected information to the search device 2. The information acquisition unit 9 may also be configured as a means for identifying location information by scanning map information. The information acquisition unit 9 may also be configured as a temperature sensor, a humidity sensor, a wind direction sensor, and an illuminance sensor for measuring temperature, humidity, wind direction, and light. The information acquisition unit 9 may also be configured as a communication interface for acquiring weather data from the Japan Meteorological Agency or private weather forecasting companies. The information acquisition unit 9 may also be configured as a body sensor worn on the body to detect body data, and this body sensor may be configured as a sensor for detecting, for example, body temperature, heart rate, blood pressure, number of steps, walking speed, and acceleration. The body sensor may also be configured to acquire biometric data of not only humans but also animals. The information acquisition unit 9 may also be configured as a device that acquires information by scanning information such as drawings or reading it from a database. In addition to these, the information acquisition unit 9 may also be configured as an odor sensor that detects odors and fragrances.

[0012] In addition, the information acquisition unit 9 may be configured as a means for acquiring sales data for each region recorded in a database of contractors who respond to plumbing problems and solve the problems by actually performing the work, and dispatch frequency data calculated based on the number of dispatches for each region, etc.

[0013] Database 3 stores various information necessary for calculating the frequency of occurrence of plumbing problems. Information necessary for calculating the frequency of occurrence of plumbing problems includes reference sales data of contractors who have been dispatched to respond to plumbing problems in each region, reference dispatch frequency data of contractors who have been dispatched to respond to plumbing problems in each region, reference refusal rate of contractors requested to respond to plumbing problems, reference population estimate data for each region, reference geographical information for each region, reference trouble type information regarding the types of trouble in each region, and reference statistical information regarding the types of building structures in each region, all of which are stored in relation to the frequency of occurrence of plumbing problems as output data.

[0014] In other words, in addition to the contractor's reference sales data and contractor's reference dispatch frequency data, database 3 stores one or more of the contractor's reference refusal rate, reference population estimate data, reference geographical information, reference trouble type information, and reference statistical information on the type of building structure, and the frequency of occurrence of plumbing troubles, all of which are linked to each other.

[0015] The search device 2 is configured with an electronic device such as a personal computer (PC), but may also be realized with any other electronic device other than a PC, such as a mobile phone, a smartphone, a tablet terminal, a wearable terminal, etc. The user can obtain a search solution by this search device 2.

[0016] 2 shows a specific example of the configuration of the search device 2. In this search device 2, a control unit 24 for controlling the entire search device 2, an operation unit 25 for inputting various control commands via operation buttons, a keyboard, etc., a communication unit 26 for performing wired or wireless communication, a discrimination unit 27 for making various decisions, and a memory unit 28, represented by a hard disk or the like, for storing programs for performing searches to be executed, are all connected to an internal bus 21. Furthermore, a display unit 23 serving as a monitor for actually displaying information is connected to this internal bus 21.

[0017] The control unit 24 is a so-called central control unit that controls each component implemented in the searching device 2 by transmitting a control signal via the internal bus 21. The control unit 24 also transmits various control commands via the internal bus 21 in response to operations via the operation unit 25.

[0018] The operation unit 25 is embodied by a keyboard or a touch panel, and an execution command for executing a program is input by the user. When the execution command is input by the user, the operation unit 25 notifies the control unit 24. Upon receiving this notification, the control unit 24 executes the desired processing operation in cooperation with the determination unit 27 and other components. The operation unit 25 may be embodied as the information acquisition unit 9 described above.

[0019] The discrimination unit 27 discriminates the search solution. When performing the discrimination operation, the discrimination unit 27 reads out various pieces of information stored in the storage unit 28 as necessary information and various pieces of information stored in the database 3. The discrimination unit 27 may be controlled by artificial intelligence. The artificial intelligence may be based on any well-known artificial intelligence technology.

[0020] The display unit 23 is configured by a graphic controller that creates a display image under the control of the control unit 24. The display unit 23 is realized by, for example, a liquid crystal display (LCD) or the like.

[0021] When the storage unit 28 is configured as a hard disk, predetermined information is written to each address and read out as necessary under the control of the control unit 24. The storage unit 28 also stores a program for carrying out the present invention. The program is read out and executed by the control unit 24.

[0022] The operation of the plumbing trouble occurrence frequency system 1 configured as described above will be described.

[0023] The plumbing trouble occurrence frequency system 1 is based on the premise that, as shown in Figure 3, three or more levels of correlation are predefined between the reference sales data of contractors dispatched to handle plumbing troubles in each region and the frequency of plumbing troubles. Contractors dispatched to handle plumbing troubles in each region are those who respond to plumbing trouble requests from residents of building structures (buildings, condominiums, detached houses, apartments, etc.) and actually perform on-site repair work. Plumbing troubles include toilet clogs, leaks, and malfunctions, kitchen and bathroom clogs, leaks, and malfunctions, and clogged bathtub drains, as well as shower and faucet leaks and malfunctions. These contractors often manage their sales by region. Individual regions can be classified into regions, prefectures, cities, wards, towns, towns, addresses, and even buildings and condominiums. Sales are managed by year, month, week, day, etc. Reference sales data for these contractors by region is first acquired as training data. Furthermore, this reference sales data may be expressed as an average value or standard deviation for a certain period, such as yearly, monthly, weekly, or daily, or may be expressed as data on fluctuation trends or fluctuation transitions.

[0024] The frequency of plumbing problems indicates how often plumbing problems are likely to occur in each region. This plumbing problem frequency can be calculated in any denominator, such as annual, monthly, weekly, daily, or five-year increments. A plumbing problem can be counted as one occurrence each time a resident of a building structure contacts a contractor to report a plumbing problem. The plumbing problem frequency can be counted by the contractor themselves and recorded in a database, allowing for subsequent retrieval. This plumbing problem frequency is organized by the region mentioned above.

[0025] In the example of Figure 3, the input data is assumed to be reference sales data P01, P02, and P03 for each region. Such reference sales data P01, P02, and P03 as input data are linked to the frequency of plumbing troubles as output.

[0026] The reference sales data P01, P02, and P03 are correlated with the plumbing trouble frequencies A to D as the output solution through three or more levels of correlation. While these trouble frequencies are shown, for example, as A = 5 times per month and B = 20 times per month, they may be expressed in any periodic unit, not limited to monthly. The reference sales data are arranged on the left side according to this correlation, and the plumbing trouble frequencies are arranged on the right side according to the correlation. The correlation indicates the degree to which the reference sales data arranged on the left side are highly correlated with each trouble frequency. In other words, the correlation is an index indicating the likelihood that each reference sales data is linked to a specific trouble frequency, and indicates the accuracy of selecting the most likely trouble frequency for each reference sales data. In the example of Figure 3, correlations w13 to w19 are shown. These w13 to w19 are shown on a 10-point scale as shown in Table 1 below. The closer to 10 points, the more closely each combination as an intermediate node is related to the frequency of trouble occurring as an output. Conversely, the closer to 1 point, the less closely each combination as an intermediate node is related to the frequency of trouble occurring as an output.

[0027] [Table 1]

[0028] The search device 2 acquires in advance three or more levels of correlation w13 to w19 as shown in Fig. 3. In other words, when determining the actual search solution, the search device 2 accumulates past data sets, including reference sales data for each region and the frequency of trouble occurrence in that case, which was adopted and evaluated, and analyzes these to create the correlations shown in Fig. 3.

[0029] For example, let's say that past regional reference sales data is shown as a line graph with a trend of fluctuations. In the case of reference sales data like this, it's assumed that the frequency of plumbing problems in that region was actually highest in A. By collecting and analyzing such data sets, the correlation with the reference sales data for each region becomes stronger.

[0030] This analysis may be performed using artificial intelligence. In such a case, for example, in the case of reference sales data P01, analysis is performed using various data on past sales and trouble occurrence frequency. If the average annual sales in region P01 is 5.6 million yen and there are many cases of trouble occurrence frequency A, the correlation degree leading to the evaluation of this trouble occurrence frequency is set higher, and if there are many cases of trouble occurrence frequency B, the correlation degree leading to the evaluation of this trouble occurrence frequency is set higher. For example, in the example of reference sales data for region P01, trouble occurrence frequency A is linked to trouble occurrence frequency C, and the correlation degree of w13, which connects previous cases to trouble occurrence frequency A, is set to 7 points, and the correlation degree of w14, which connects previous cases to trouble occurrence frequency C, is set to 2 points.

[0031] The correlation shown in Fig. 3 may be configured by nodes of a neural network in artificial intelligence. That is, the weighting coefficients for the outputs of the nodes of this neural network correspond to the correlation described above. Furthermore, the correlation is not limited to a neural network, and may be configured by any decision-making factors that constitute artificial intelligence.

[0032] In such a case, as shown in Figure 4, reference sales data for each region may be input as input data, and trouble occurrence frequency may be output as output data, with at least one or more hidden layers provided between the input node and the output node for machine learning. The above-mentioned correlation degree may be set in either or both of the input node and the hidden layer node, which then serves as a weight for each node, and output selection is based on this. If this correlation degree exceeds a certain threshold, that output may be selected.

[0033] This correlation is what is called "trained data" in artificial intelligence. After creating this trained data through a dataset of previous reference sales data for each region and the frequency of occurrence of plumbing problems, the trained data described above will be used to search for the frequency of occurrence of problems when actually determining the frequency of new problems. These datasets may be created by reading them from a database managed by the contractor.

[0034] When searching for a new frequency of trouble occurrence, input the region you want to search for. Since past sales data is linked to each region and stored, when you input a region, you can obtain it by reading the sales data linked to it.

[0035] Next, this read-out sales data is compared with the reference sales data. In such cases, the correlation degrees shown in Figure 3 (Table 1) obtained in advance are referenced. For example, if the newly obtained sales data is identical to or similar to P02, the trouble occurrence frequency B is associated with the correlation degree w15, and the trouble occurrence frequency C is associated with the correlation degree w16. In such a case, the trouble occurrence frequency B, which has the highest correlation degree, is selected as the optimal solution. However, it is not essential to select the one with the highest correlation degree as the optimal solution; the trouble occurrence frequency C, which has a low correlation degree but is recognized as being correlated, may also be selected as the optimal solution. Furthermore, it is of course possible to select an output solution other than this that does not have a connecting arrow, and any other priority order may be used as long as it is based on the correlation degree.

[0036] Incidentally, when comparing sales data and reference sales data, if these data are expressed as average sales for a certain period, whether they are identical or similar may be determined based on whether the average sales is within a range of ±10%. Also, if the sales data is shown as a time-series trend graph, the determination may be based on the similarity of the trends.

[0037] In this way, the most suitable trouble occurrence frequency can be found from newly acquired sales data and displayed to the user. By looking at the search results, it is possible to determine in advance what kind of trouble occurrence frequency is likely to occur in that area in the future, and to consider the allocation of workers in each area.

[0038] The example in Figure 5 shows how a correlation is formed between the combination of reference sales data and reference refusal rate. The reference refusal rate is the probability that an actual request for response to a plumbing problem is made to a contractor by a resident of a building structure, and the contractor is unable to accept the request and has to decline. This refusal rate is expressed as the number of refusals relative to the number of response requests. The number of response requests and the number of refusals are managed by the contractor for each region in database 3. For the region for which you want to know the actual refusal rate, you can obtain the refusal rate by reading the number of refusals relative to the number of response requests from database 3.

[0039] The frequency of occurrence of plumbing troubles depends not only on sales in the area, but also on the refusal rate, since if there are too many requests for dispatch, there will be many cases where services will be turned down. For this reason, by combining the reference sales data and the reference refusal rate with the learning data, the frequency of occurrence of troubles can be determined with higher accuracy. For this reason, the reference sales data and the reference refusal rate are combined to form the above-mentioned correlation.

[0040] In the example of Figure 5, the input data may be, for example, reference sales data P01 to P03 and reference refusal rates P14 to P17. The intermediate nodes shown in Figure 5 are formed by combining the reference sales data as input data with the reference refusal rates. Each intermediate node is further connected to an output. In this output, the frequency of trouble occurrence is displayed as an output solution.

[0041] Each combination (intermediate node) of reference sales data and reference refusal rates is correlated with the trouble occurrence frequency as the output solution through three or more levels of correlation. The reference sales data and reference refusal rates are arranged on the left side via this correlation, and the trouble occurrence frequency is arranged on the right side via this correlation. The correlation indicates the degree to which the trouble occurrence frequency is highly correlated with the reference sales data and reference refusal rates arranged on the left side. In other words, this correlation is an indicator of the likelihood that each reference sales data and reference refusal rate will be linked to what trouble occurrence frequency, and indicates the accuracy of selecting the most likely trouble occurrence frequency from the reference sales data and reference refusal rate. Therefore, the optimal trouble occurrence frequency will be searched for by combining these reference sales data and reference refusal rates.

[0042] In the example of Figure 5, w13 to w22 are shown as the degrees of association. These w13 to w22 are shown on a 10-point scale as shown in Table 1, with the closer to 10 points the higher the degree of association between each combination as an intermediate node and the output, and conversely, the closer to 1 point the lower the degree of association between each combination as an intermediate node and the output.

[0043] The search device 2 acquires in advance three or more levels of correlation w13 to w22 as shown in Fig. 5. In other words, when determining an actual search solution, the search device 2 accumulates past data on which reference sales data and reference refusal rates, as well as the frequency of trouble occurrence in each case, are appropriate, and by analyzing these, creates the correlations shown in Fig. 5.

[0044] This analysis may be performed using artificial intelligence. In such a case, for example, if the reference sales data P01 has a reference refusal rate P16, the trouble occurrence frequency is analyzed from past data. If there are many cases of trouble occurrence frequency A, the correlation degree leading to trouble occurrence frequency A is set higher. If there are many cases of trouble occurrence frequency B and few cases of trouble occurrence frequency A, the correlation degree leading to trouble occurrence frequency B is set higher and the correlation degree leading to trouble occurrence frequency A is set lower. For example, in the example of intermediate node 61a, which is linked to the outputs of trouble occurrence frequency A and trouble occurrence frequency B, the correlation degree of w13, which connects from previous cases to trouble occurrence frequency A, is set to 7 points, and the correlation degree of w14, which connects to trouble occurrence frequency B, is set to 2 points.

[0045] The correlation shown in Figure 5 may be configured by nodes of a neural network in artificial intelligence. That is, the weighting coefficients for the output of the nodes of this neural network correspond to the correlation described above. Furthermore, it is not limited to a neural network, and may be configured by any decision-making factors that constitute artificial intelligence. Other configurations related to artificial intelligence are the same as those described in Figure 4.

[0046] In the example of correlation degrees shown in Figure 5, node 61b is a node for the combination of reference sales data P01 and reference refusal rate P14, and the correlation degree for trouble occurrence frequency C is w15 and the correlation degree for trouble occurrence frequency E is w16. Node 61c is a node for the combination of reference refusal rates P15 and P17 for reference sales data P02, and the correlation degree for trouble occurrence frequency B is w17 and the correlation degree for trouble occurrence frequency D is w18.

[0047] This correlation becomes what is called learned data in artificial intelligence. After creating this learned data, the above-mentioned learned data will be used when actually determining the frequency of trouble occurrence. In such cases, the region for which the frequency of trouble occurrence is to be determined is entered in the same way. Then, the sales data and error rate organized for each region in Database 3 are obtained.

[0048] In this way, the optimal trouble occurrence frequency is searched for based on the newly acquired sales data and error rate. In this case, the previously acquired correlation shown in Figure 5 (Table 1) is referenced. For example, if the newly acquired sales data is identical to or similar to P02 and the error rate is identical to or similar to P17, node 61d is associated via correlation, and this node 61d is associated with trouble occurrence frequency C at w19 and trouble occurrence frequency D at correlation w20. In this case, trouble occurrence frequency C, which has the highest correlation, is selected as the optimal solution. However, it is not necessary to select the one with the highest correlation as the optimal solution; trouble occurrence frequency D, which has a low correlation but is recognized as being correlated, may also be selected as the optimal solution. Of course, other output solutions not connected by arrows may also be selected, and any other priority may be selected based on correlation.

[0049] Table 2 below shows examples of the degrees of association w1 to w12 extending from the input.

[0050] [Table 2]

[0051] The intermediate node 61 may be selected based on the degrees of association w1 to w12 extending from this input. In other words, the greater the degrees of association w1 to w12, the heavier the weighting in selecting the intermediate node 61. However, the degrees of association w1 to w12 may all have the same value, and the weighting in selecting the intermediate node 61 may all be the same.

[0052] Figure 6 shows an example in which, in addition to the reference sales data described above, reference population estimate data is used in place of the reference refusal rate described above, and three or more levels of correlation are set between the combination and the frequency of trouble occurrence.

[0053] This reference population estimate data, which is added as an explanatory variable in place of the reference refusal rate, shows the population estimate for the region. It may also include a population pyramid (a diagram showing the distribution of the population by age group and gender), time-series trends, the number of people moving in and out of the region, the number of households moving in and out, and occupational breakdowns for each population. The frequency of trouble occurrences is affected by such population estimates in addition to sales data. The larger the elderly population, the more likely it is that services will be unable to deal with clogged toilets, which may increase the number of times a service is called out. Furthermore, the greater the positive difference between the number of people moving in and the number of people moving out, the greater the population is growing, and the corresponding frequency of trouble occurrences is likely to increase. Such reference population estimate data is managed in Database 3 for each region.

[0054] Since such population estimates also affect the frequency of trouble occurrence, the accuracy of the determination can be improved by combining them with reference sales data and determining the frequency of trouble occurrence through the degree of correlation.

[0055] In the example of Figure 6, the input data is assumed to be, for example, reference sales data P01 to P03 and reference population projection data P18 to P21. The intermediate nodes shown in Figure 6 are created by combining the reference sales data as input data with the reference population projection data. Each intermediate node is further connected to an output. In this output, the frequency of trouble occurrence is displayed as an output solution.

[0056] Each combination (intermediate node) of reference sales data and reference population estimate data is correlated with each other through three or more levels of correlation to the trouble occurrence frequency as the output solution. The reference sales data and reference population estimate data are arranged on the left side via this correlation, and the trouble occurrence frequency is arranged on the right side via this correlation. The correlation indicates the degree to which the trouble occurrence frequency is highly related to the reference sales data and reference population estimate data arranged on the left side. In other words, this correlation is an index that indicates the likelihood that each reference sales data and reference population estimate data will be linked to what kind of trouble occurrence frequency, and indicates the accuracy of selecting the most likely trouble occurrence frequency from the reference sales data and reference population estimate data.

[0057] The search device 2 acquires in advance three or more levels of correlation w13 to w22 as shown in Fig. 6. That is, when determining an actual search solution, the search device 2 accumulates reference sales data, reference population estimate data obtained when acquiring the reference sales data, and past data on which of the trouble occurrence frequencies was most suitable, and analyzes these to create the correlations shown in Fig. 7.

[0058] This analysis may be performed using artificial intelligence. In such a case, for example, when the reference sales data is P01 and the reference population projection data is P20, the trouble occurrence frequency is analyzed from past data. If there are many cases of trouble occurrence frequency A, the correlation degree connecting this trouble occurrence frequency to A is set higher. If there are many cases of trouble occurrence frequency B and few cases of trouble occurrence frequency A, the correlation degree connecting the trouble occurrence frequency to B is set higher and the correlation degree connecting the trouble occurrence frequency to A is set lower. For example, in the example of intermediate node 61a, which is linked to the outputs of trouble occurrence frequency A and trouble occurrence frequency B, the correlation degree of w13 connecting from previous cases to trouble occurrence frequency A is set to 7 points, and the correlation degree of w14 connecting to trouble occurrence frequency B is set to 2 points.

[0059] The correlation shown in Figure 6 may be configured by nodes of a neural network in artificial intelligence. That is, the weighting coefficients for the output of the nodes of this neural network correspond to the correlation described above. Furthermore, it is not limited to a neural network, and may be configured by any decision-making factors that constitute artificial intelligence. Other configurations related to artificial intelligence are the same as those described in Figure 4.

[0060] In the example of correlation degrees shown in Figure 6, node 61b is a node that combines reference sales data P01 and reference population estimate data P18, and the correlation degree for trouble occurrence frequency C is w15 and the correlation degree for trouble occurrence frequency E is w16. Node 61c is a node that combines reference population estimate data P19 and P21 with reference sales data P02, and the correlation degree for trouble occurrence frequency B is w17 and the correlation degree for trouble occurrence frequency D is w18.

[0061] This correlation is what is called "trained data" in artificial intelligence. After creating this trained data, the trained data will be used when searching for the frequency of trouble occurrence. In such cases, by actually entering the region for which the frequency of trouble occurrence is to be determined, sales data and population estimate data for that region will be obtained.

[0062] In this way, the optimal trouble occurrence frequency is searched for based on the newly acquired sales data and population projection data. In this case, the previously acquired correlation shown in Figure 6 (Table 1) is referenced. For example, if the newly acquired sales data is identical to or similar to P02 and the population projection data is identical to or similar to P21, node 61d is associated via correlation, and this node 61d is associated with trouble occurrence frequency C at w19 and trouble occurrence frequency D at correlation w20. In this case, the trouble occurrence frequency C with the highest correlation is selected as the optimal solution. However, it is not necessary to select the one with the highest correlation as the optimal solution; trouble occurrence frequency D with a low correlation but a recognized correlation may also be selected as the optimal solution. Of course, other output solutions without connected arrows may also be selected, and any other priority order may be used as long as it is based on correlation.

[0063] Figure 7 shows an example in which, in addition to the reference sales data described above, a combination of reference geographical information is set instead of the reference refusal rate described above, and three or more levels of correlation are set between the frequency of trouble occurrence for that combination.

[0064] This reference geographical information, which is added as an explanatory variable instead of the reference refusal rate, indicates all geographical information for the region, such as the presence or absence of rivers and sea, their locations, distances, areas, meters above sea level, contour information, road information, the relative positional relationship of building structures to rivers, etc. Such reference geographical information is managed in database 3 for each region.

[0065] Such geographical information also influences the frequency of plumbing problems. Proximity to a river can increase the likelihood of plumbing problems occurring, so by combining this with reference sales data and determining the frequency of problems through correlation, the accuracy of the determination can be improved.

[0066] In the example of Figure 7, the input data is assumed to be, for example, reference sales data P01 to P03 and reference geographical information P18 to P21. The intermediate nodes shown in Figure 7 are formed by combining the reference sales data as input data with the reference geographical information. Each intermediate node is further connected to an output. In this output, the frequency of trouble occurrence is displayed as an output solution.

[0067] Each combination (intermediate node) of reference sales data and reference geographic information is correlated with the trouble occurrence frequency as the output solution through three or more levels of correlation. The reference sales data and reference geographic information are arranged on the left side via this correlation, and the trouble occurrence frequency is arranged on the right side via this correlation. The correlation indicates the degree to which the trouble occurrence frequency is highly related to the reference sales data and reference geographic information arranged on the left side. In other words, this correlation is an index that indicates the likelihood that each reference sales data and reference geographic information will be linked to what trouble occurrence frequency, and indicates the accuracy of selecting the most likely trouble occurrence frequency from the reference sales data and reference geographic information.

[0068] The search device 2 acquires in advance three or more levels of correlation w13 to w22 as shown in Fig. 7. In other words, when determining an actual search solution, the search device 2 accumulates reference sales data, reference geographical information obtained when acquiring the reference sales data, and past data on which of the trouble occurrence frequencies was most suitable, and analyzes these to create the correlations shown in Fig. 7.

[0069] This analysis may be performed using artificial intelligence. In such a case, for example, when reference sales data P01 and reference geographic information P20 are used, the trouble occurrence frequency is analyzed from past data. If there are many cases of trouble occurrence frequency A, the correlation degree connecting this trouble occurrence frequency to A is set higher. If there are many cases of trouble occurrence frequency B and few cases of trouble occurrence frequency A, the correlation degree connecting the trouble occurrence frequency to B is set higher and the correlation degree connecting the trouble occurrence frequency to A is set lower. For example, in the example of intermediate node 61a, which is linked to the outputs of trouble occurrence frequency A and trouble occurrence frequency B, the correlation degree of w13 connecting from previous cases to trouble occurrence frequency A is set to 7 points, and the correlation degree of w14 connecting to trouble occurrence frequency B is set to 2 points.

[0070] The correlation shown in Fig. 7 may be configured by nodes of a neural network in artificial intelligence. That is, the weighting coefficient for the output of the node of this neural network corresponds to the correlation described above. Furthermore, it is not limited to a neural network, and may be configured by any decision-making factor that constitutes artificial intelligence. Other configurations related to artificial intelligence are the same as those described in Fig. 4.

[0071] 7, node 61b is a node that combines reference sales data P01 with reference geographical information P18, and the degree of association for trouble occurrence frequency C is w15 and the degree of association for trouble occurrence frequency E is w16. Node 61c is a node that combines reference geographical information P19 and P21 with reference sales data P02, and the degree of association for trouble occurrence frequency B is w17 and the degree of association for trouble occurrence frequency D is w18.

[0072] This correlation is what is called "trained data" in artificial intelligence. After creating this trained data, the trained data will be used when searching for trouble occurrence frequency. In such cases, by actually entering the region for which the trouble occurrence frequency is to be determined, sales data and geographical information for that region will be obtained.

[0073] In this way, the optimal trouble occurrence frequency is searched for based on the newly acquired sales data and geographical information. In this case, the previously acquired correlation shown in Figure 7 (Table 1) is referenced. For example, if the newly acquired sales data is identical to or similar to P02 and the geographical information is identical to or similar to P21, node 61d is associated via correlation, and this node 61d is associated with trouble occurrence frequency C at w19 and trouble occurrence frequency D at correlation w20. In this case, the trouble occurrence frequency C with the highest correlation is selected as the optimal solution. However, it is not necessary to select the one with the highest correlation as the optimal solution; trouble occurrence frequency D with a low correlation but a recognized correlation may be selected as the optimal solution. Furthermore, it is of course possible to select an output solution without an arrow connecting it to the others, and any other priority may be selected based on the correlation.

[0074] Figure 8 shows an example in which, in addition to the reference sales data described above, a combination of reference trouble type information instead of the reference refusal rate described above is set, and three or more levels of correlation are set between the combination and the trouble occurrence frequency.

[0075] This reference trouble type information, which is added as an explanatory variable instead of the reference refusal rate, shows information on all types of trouble in the area. These trouble types are classified into toilet clogs, water leaks, malfunctions, kitchen and bathroom clogs, water leaks, malfunctions, bathtub drain clogs, shower and faucet leaks, malfunctions, etc. Such reference trouble type information is managed in database 3 for each area.

[0076] This type of trouble information also affects the frequency of plumbing troubles. The frequency of troubles can vary depending on whether there are many water pipe leaks or clogged drains, so by combining this with reference sales data and determining the frequency of troubles through correlation, the accuracy of the determination can be improved.

[0077] In the example of Figure 8, the input data is assumed to be, for example, reference sales data P01 to P03 and reference trouble type information P18 to P21. The intermediate nodes shown in Figure 8 are formed by combining the reference sales data as input data with the reference trouble type information. Each intermediate node is further connected to an output. In this output, the frequency of trouble occurrence is displayed as an output solution.

[0078] Each combination (intermediate node) of reference sales data and reference trouble type information is correlated with the trouble occurrence frequency as the output solution through three or more levels of correlation. The reference sales data and reference trouble type information are arranged on the left side via this correlation, and the trouble occurrence frequency is arranged on the right side via this correlation. The correlation indicates the degree to which the trouble occurrence frequency is highly correlated with the reference sales data and reference trouble type information arranged on the left side. In other words, this correlation is an index that indicates the likelihood that each reference sales data and reference trouble type information will be linked to what trouble occurrence frequency, and indicates the accuracy of selecting the most likely trouble occurrence frequency from the reference sales data and reference trouble type information.

[0079] The search device 2 acquires in advance three or more levels of correlation w13 to w22 as shown in Fig. 8. That is, when determining an actual search solution, the search device 2 accumulates reference sales data, reference trouble type information obtained when acquiring the reference sales data, and past data on which trouble occurrence frequency was most suitable in that case, and analyzes these to create the correlations shown in Fig. 8.

[0080] This analysis may be performed using artificial intelligence. In such a case, for example, when reference sales data P01 and reference trouble type information P20 are used, the trouble occurrence frequency is analyzed from past data. If there are many cases of trouble occurrence frequency A, the correlation degree connecting this trouble occurrence frequency to A is set higher. If there are many cases of trouble occurrence frequency B and few cases of trouble occurrence frequency A, the correlation degree connecting the trouble occurrence frequency to B is set higher and the correlation degree connecting the trouble occurrence frequency to A is set lower. For example, in the example of intermediate node 61a, the outputs of trouble occurrence frequency A and trouble occurrence frequency B are linked, but the correlation degree of w13 connecting from previous cases to trouble occurrence frequency A is set to 7 points, and the correlation degree of w14 connecting to trouble occurrence frequency B is set to 2 points.

[0081] The correlation shown in Fig. 8 may be configured by nodes of a neural network in artificial intelligence. That is, the weighting coefficients for the output of the nodes of this neural network correspond to the correlations described above. Furthermore, it is not limited to a neural network, and may be configured by any decision-making factors that constitute artificial intelligence. Other configurations related to artificial intelligence are the same as those described in Fig. 4.

[0082] 8, node 61b is a node representing the combination of reference sales data P01 and reference trouble type information P18, with the degree of association for trouble occurrence frequency C being w15 and the degree of association for trouble occurrence frequency E being w16. Node 61c is a node representing the combination of reference trouble type information P19 and P21 with reference sales data P02, with the degree of association for trouble occurrence frequency B being w17 and the degree of association for trouble occurrence frequency D being w18.

[0083] This correlation becomes what is called "trained data" in artificial intelligence. After creating this trained data, the trained data will be used when searching for trouble occurrence frequency. In such cases, by actually entering the region for which the trouble occurrence frequency is to be determined, sales data and trouble type information for that region will be obtained.

[0084] In this way, the optimal trouble occurrence frequency is searched for based on the newly acquired sales data and trouble type information. In this case, the previously acquired correlation shown in FIG. 8 (Table 1) is referenced. For example, if the newly acquired sales data is identical to or similar to P02 and the trouble type information is identical to or similar to P21, node 61d is associated via correlation, and this node 61d is associated with trouble occurrence frequency C at w19 and trouble occurrence frequency D at correlation w20. In this case, the trouble occurrence frequency C with the highest correlation is selected as the optimal solution. However, it is not necessary to select the one with the highest correlation as the optimal solution; trouble occurrence frequency D with a low correlation but a recognized correlation may also be selected as the optimal solution. Of course, other output solutions without connected arrows may also be selected, and any other priority order may be used as long as it is based on correlation.

[0085] Figure 9 shows an example in which, in addition to the reference sales data described above, a combination of reference statistical information instead of the reference refusal rate described above is set, and three or more levels of correlation are set between the frequency of trouble occurrence for that combination.

[0086] This reference statistical information, which is added as an explanatory variable instead of the reference refusal rate, is statistical information on the types of building structures in the region. The types of building structures include broad categories such as buildings, condominiums, detached houses, apartments, etc., as well as statistical analysis of the age and construction method (light steel frame, heavy steel frame, reinforced concrete, wood, etc.) of each building structure. The proportions of these types, construction methods, and ages are statistically analyzed to facilitate comparison and analysis between regions. Such reference statistical information is managed in database 3 for each region. This statistical information may also be composed of statistical information on the age of building structures in the region.

[0087] Such statistical information also influences the frequency of plumbing problems. As the age of a building increases, there are cases where water pipes are more likely to leak and drains are more likely to be clogged, so by combining this with reference sales data and determining the frequency of problems through correlation, the accuracy of the determination can be improved.

[0088] In the example of Figure 9, the input data is assumed to be, for example, reference sales data P01 to P03 and reference statistical information P18 to P21. The intermediate nodes shown in Figure 9 are formed by combining the reference sales data as input data with the reference statistical information. Each intermediate node is further connected to an output. In this output, the frequency of trouble occurrence is displayed as an output solution.

[0089] Each combination (intermediate node) of reference sales data and reference statistical information is correlated with the trouble occurrence frequency as the output solution through three or more levels of correlation. The reference sales data and reference statistical information are arranged on the left side via this correlation, and the trouble occurrence frequency is arranged on the right side via this correlation. The correlation indicates the degree to which the trouble occurrence frequency is highly related to the reference sales data and reference statistical information arranged on the left side. In other words, this correlation is an index that indicates the likelihood that each reference sales data and reference statistical information will be linked to what trouble occurrence frequency, and indicates the accuracy of selecting the most likely trouble occurrence frequency from the reference sales data and reference statistical information.

[0090] The search device 2 acquires in advance three or more levels of correlation w13 to w22 as shown in Fig. 9. In other words, when determining an actual search solution, the search device 2 accumulates reference sales data, reference statistical information obtained when acquiring the reference sales data, and past data on which of the trouble occurrence frequencies was most suitable, and analyzes these to create the correlations shown in Fig. 9.

[0091] This analysis may be performed using artificial intelligence. In such a case, for example, when the reference sales data is P01 and the reference statistical information is P20, the trouble occurrence frequency is analyzed from past data. If there are many cases of trouble occurrence frequency A, the correlation degree connecting this trouble occurrence frequency to A is set higher. If there are many cases of trouble occurrence frequency B and few cases of trouble occurrence frequency A, the correlation degree connecting the trouble occurrence frequency to B is set higher and the correlation degree connecting the trouble occurrence frequency to A is set lower. For example, in the example of intermediate node 61a, the outputs of trouble occurrence frequency A and trouble occurrence frequency B are linked, but the correlation degree of w13 connecting from previous cases to trouble occurrence frequency A is set to 7 points, and the correlation degree of w14 connecting to trouble occurrence frequency B is set to 2 points.

[0092] The correlation shown in Fig. 9 may be configured by nodes of a neural network in artificial intelligence. That is, the weighting coefficient for the output of the node of this neural network corresponds to the correlation described above. Furthermore, it is not limited to a neural network, and may be configured by any decision-making factor that constitutes artificial intelligence. Other configurations related to artificial intelligence are the same as those described in Fig. 4.

[0093] 9, node 61b is a node representing a combination of reference sales data P01 and reference statistical information P18, with the degree of association for trouble occurrence frequency C being w15 and the degree of association for trouble occurrence frequency E being w16. Node 61c is a node representing a combination of reference statistical information P19 and P21 for reference sales data P02, with the degree of association for trouble occurrence frequency B being w17 and the degree of association for trouble occurrence frequency D being w18.

[0094] This correlation is what is called "trained data" in artificial intelligence. After creating this trained data, the trained data will be used when searching for the frequency of trouble occurrence. In such cases, by actually entering the region for which the frequency of trouble occurrence is to be determined, sales data and statistical information for that region will be obtained.

[0095] In this way, the optimal trouble occurrence frequency is searched for based on the newly acquired sales data and statistical information. In this case, the previously acquired correlation shown in Figure 9 (Table 1) is referenced. For example, if the newly acquired sales data is identical to or similar to P02 and the statistical information is identical to or similar to P21, node 61d is associated via the correlation, and this node 61d is associated with trouble occurrence frequency C at w19 and trouble occurrence frequency D at a correlation of w20. In this case, the trouble occurrence frequency C with the highest correlation is selected as the optimal solution. However, it is not necessary to select the one with the highest correlation as the optimal solution; trouble occurrence frequency D with a low correlation but a recognized correlation may also be selected as the optimal solution. Of course, other output solutions without connected arrows may also be selected, and any other priority order may be used as long as it is based on the correlation.

[0096] Furthermore, as an alternative to the above-mentioned reference information (reference refusal rate, reference population estimate data, reference geographical information, reference problem type information, reference statistical information, etc.), reference time information regarding the time when the reference sales data was acquired may be used. The reference time information here refers to the time when the reference sales data was acquired, and is made up of all data indicating the time, such as month, week, day, season, etc. There are also times when plumbing problems are more likely to occur, which affects the frequency of plumbing problems, so by taking this into account in the judgment, a more accurate solution search can be achieved.

[0097] In such a case, three or more levels of correlation between the frequency of occurrence of plumbing problems and a combination of reference sales data of contractors who have responded to plumbing problems in each region in the past and reference timing information regarding the time when the reference sales data was acquired is acquired in advance. Then, area-specific information for identifying the building structure for which the frequency of occurrence of plumbing problems is to be predicted or the region in which it is located and timing information regarding the time when the area-specific information was acquired are acquired. Next, the frequency of occurrence of plumbing problems is searched for based on the correlations shown in Figures 5 to 9, using the reference sales data corresponding to the past sales data of the region in the acquired area-specific information and the reference timing information corresponding to the acquired timing information.

[0098] Furthermore, as an alternative to the above-mentioned reference information (reference refusal rate, reference population estimate data, reference geographical information, reference trouble type information, reference statistical information, etc.), reference weather information relating to the weather at the time when the area-specific information was obtained may be used. Reference weather information here refers to all information relating to the weather at the time when the reference sales data was obtained, and is made up of all data relating to climate, temperature, humidity, weather, wind direction, wind speed, thunderstorms, typhoons, droughts, etc. Weather is also a factor that affects plumbing troubles, so it has been added as an explanatory variable.

[0099] In such a case, three or more levels of correlation between the frequency of occurrence of plumbing problems and a combination of reference sales data of contractors who have responded to plumbing problems in each region in the past and reference weather information regarding the weather at the time the reference sales data was obtained is acquired in advance. Then, region-specific information for identifying the building structure for which the frequency of occurrence of plumbing problems is to be predicted or the region in which it is located, and weather information regarding the weather at the time the region-specific information was obtained are acquired. Next, the frequency of occurrence of plumbing problems is searched for based on the correlations shown in Figures 5 to 9, using the reference sales data corresponding to the past sales data of the region in the acquired region-specific information and the reference weather information corresponding to the acquired weather information.

[0100] In the above-mentioned correlation degree, the correlation degree is expressed on a 10-point scale, but it is not limited to this and may be expressed on a scale of 3 or more, and conversely, if it is 3 or more, it may be expressed on a scale of 100 or 1000. On the other hand, this correlation degree does not include a 2-point scale, that is, a scale expressed by either 1 or 0, indicating whether or not there is a correlation between the two.

[0101] According to the present invention having the above-mentioned configuration, anyone can easily determine and search for trouble occurrence frequency, even without special skills or experience. Furthermore, according to the present invention, it is possible to determine the search solution with higher accuracy than a human. Furthermore, by configuring the above-mentioned correlation using artificial intelligence (neural network, etc.), it is possible to further improve the determination accuracy by learning this.

[0102] In addition, since there are many cases where the above-mentioned input data and output data do not exist exactly the same during the learning process, the input data and output data may be classified by type. In other words, the information P01, P02, ..., P15, 16, ... that constitutes the input data may be classified according to classification criteria previously determined by the system or user depending on the content of the information, and a data set may be created using the classified input data and output data, and learning may be performed.

[0103] The correlation degree described above is based on an example in which the correlation degree is composed of a combination of reference sales data and any one of a contractor's reference refusal rate, reference population estimate data, reference geographical information, reference trouble type information, and reference statistical information related to building structure types, but is not limited to this. That is, the correlation degree may be composed of a combination of reference sales data and any two or more of a contractor's reference refusal rate, reference population estimate data, reference geographical information, reference trouble type information, and reference statistical information related to building structure types. Furthermore, the correlation degree may be formed by combining reference sales data or, in addition to reference sales data, any one or more of a contractor's reference refusal rate, reference population estimate data, reference geographical information, reference trouble type information, and reference statistical information related to building structure types, with other factors added to the combination.

[0104] In either case, data is input according to the reference information for the correlation, and the frequency of occurrence of the trouble is calculated using the correlation.

[0105] Furthermore, the present invention determines the frequency of trouble occurrence based on the correlation between a combination of two or more types of information, reference information U and reference information V, as shown in Fig. 10. This reference information Y is reference sales data, and reference information V is any one of the contractor's reference refusal rate, reference population estimate data, reference geographical information, reference trouble type information, and reference statistical information on the type of building structure.

[0106] 10, the output obtained for the reference information U may be used as input data as it is, and may be associated with the output (frequency of trouble occurrence) via an intermediate node 61 in combination with the reference information V. For example, after an output solution is obtained for the reference information U (reference sales data) as shown in FIG. 3, this may be used as input as it is, and the degree of association with other reference information V may be used to search for the output (frequency of trouble occurrence).

[0107] Furthermore, according to the present invention, instead of obtaining the frequency of trouble occurrence as an output solution, it is also possible to issue warning information such as a warning or alarm based on the frequency of trouble occurrence. The higher the frequency of trouble occurrence, the more attention-grabbing the warning information becomes. This allows for efficient warning to the outside world that students are in a dangerous situation.

[0108] Furthermore, the present invention is characterized in that an optimal solution is searched for through correlation levels set to three or more levels. The correlation level can be expressed, for example, by a numerical value from 0 to 100%, in addition to the above-mentioned 10 levels, but is not limited to this and may be configured in any level as long as it can be expressed by a numerical value of three or more levels.

[0109] By determining the most likely frequency of occurrence of a problem based on the correlation expressed in three or more levels, it becomes possible to search and display solutions in descending order of correlation when multiple potential solutions are considered. By displaying solutions to the user in descending order of correlation, it becomes possible to prioritize the display of more likely solutions.

[0110] In addition, according to the present invention, it is possible to judge without overlooking even a discrimination result with an extremely low output, such as a correlation degree of 1%, and it is possible to alert the user that even a discrimination result with an extremely low correlation degree is connected as a slight sign, and that it may be useful as a discrimination result once in tens or hundreds of times.

[0111] Furthermore, according to the present invention, by performing a search based on such three or more levels of correlation, there is an advantage in that the search policy can be determined by how the threshold is set. A low threshold can detect even cases with a correlation of 1% without omission, but the possibility of detecting a more appropriate discrimination result is low and a lot of noise may be picked up. On the other hand, a high threshold can detect the optimal search solution with a high probability, but it may miss a suitable solution that usually has a low correlation and is ignored, but appears once in tens or hundreds of times. The emphasis can be decided based on the user's or system's perspective, and it is possible to increase the degree of freedom in selecting the points to be emphasized.

[0112] Furthermore, in the present invention, the correlation degree may be updated. This update may be made to reflect information provided via a public communication network such as the Internet. Furthermore, when reference information such as reference sales data is acquired and knowledge, information, and data regarding the frequency of trouble occurrence for the reference information is acquired, the correlation degree may be increased or decreased accordingly.

[0113] In such cases, cases of actual occurrences of each reference information, including the reference sales data, and the results of judgments of the risk levels and signs are collected, and the correlation level is increased or decreased according to the number of cases. At this time, the reference statistical information regarding the above-mentioned sales data, refusal rates, population estimate data, geographical information, trouble type information, and building structure types may be acquired, and updates may be made based on these when judgments are made.

[0114] In other words, this update corresponds to learning in artificial intelligence. Because new data is acquired and reflected in the learned data, it can be considered a learning process.

[0115] Furthermore, the degree of association may be updated not only based on information obtainable from public communication networks, but also manually or automatically by the system or user based on the contents of research data and papers by experts, academic presentations, newspaper articles, books, etc. Artificial intelligence may be utilized in these update processes.

[0116] Furthermore, the process of initially creating a trained model and the above-mentioned updates may use not only supervised learning, but also unsupervised learning, deep learning, reinforcement learning, etc. In the case of unsupervised learning, instead of reading and learning a data set of input data and output data, information corresponding to the input data may be read and learned, and then the correlation related to the output data may be self-formed from the information.

[0117] Second embodiment The second embodiment will be described below. In carrying out this second embodiment, the trouble occurrence frequency prediction system 1, information acquisition unit 9, search device 2, and database 3 used in the first embodiment are used in the same manner. The explanation of each of these components will be omitted below by quoting the explanation of the first embodiment.

[0118] In the second embodiment, a reference refusal rate and a data set of occurrence frequencies of plumbing problems are learned.

[0119] In the example of Fig. 11, the input data are the reference refusal rates P01, P02, and P03 for each region. These reference refusal rates P01, P02, and P03 as input data are linked to the frequency of occurrence of plumbing problems as output.

[0120] The reference refusal rates P01, P02, and P03 are correlated with the plumbing trouble occurrence frequencies A to D as the output solution through three or more levels of correlation. The reference refusal rates are arranged on the left side via this correlation, and the plumbing trouble occurrence frequencies are arranged on the right side via correlation. The correlation indicates the degree to which the reference refusal rates arranged on the left side are highly correlated with which trouble occurrence frequency. In other words, this correlation is an index that indicates the likelihood that each reference refusal rate is linked to which trouble occurrence frequency, and indicates the accuracy in selecting the most likely trouble occurrence frequency for each reference refusal rate. In the example of Figure 11, correlations w13 to w19 are shown. These w13 to w19 are shown on a 10-point scale as shown in Table 1 below. The closer to 10 points, the more closely each combination as an intermediate node is related to the frequency of trouble occurring as an output. Conversely, the closer to 1 point, the less closely each combination as an intermediate node is related to the frequency of trouble occurring as an output.

[0121] The search device 2 acquires in advance three or more levels of correlation w13 to w19 as shown in Fig. 11. In other words, the search device 2 accumulates past data sets on which of the reference refusal rates for each region and the trouble occurrence frequency in that case was adopted and evaluated when determining the actual search solution, and creates the correlations shown in Fig. 11 by analyzing these data sets.

[0122] The correlations shown in Fig. 11 may be configured by nodes of a neural network in artificial intelligence. That is, the weighting coefficients for the outputs of the nodes of this neural network correspond to the correlations described above. Furthermore, the correlations are not limited to neural networks, and may be configured by any decision-making factors that constitute artificial intelligence.

[0123] In such a case, as shown in Figure 12, the reference refusal rate for each region is input as input data, the trouble occurrence frequency is output as output data, and at least one hidden layer is provided between the input node and the output node, and machine learning may be performed. The above-mentioned correlation degree is set in either one or both of the input node and the hidden layer node, and this becomes the weighting of each node, and output selection is performed based on this. Then, if this correlation degree exceeds a certain threshold, that output may be selected.

[0124] This correlation is what is called "learned data" in artificial intelligence. After creating this learned data through a data set of previous reference refusal rates for each region and the frequency of occurrence of plumbing problems, the learned data will be used to search for the frequency of occurrence of problems when actually determining the frequency of new problems. These data sets may be created by reading them from a database managed by the contractor.

[0125] When searching for a new frequency of trouble occurrence, input of the area you want to search is accepted. Since the past refusal rate is linked and stored for each area, when the area is input, the refusal rate linked to it can be obtained by reading it.

[0126] Next, this read-out refusal rate is compared with the reference refusal rate. In such cases, the correlation degree shown in Figure 11 (Table 1) obtained in advance is referenced. For example, if the newly obtained refusal rate is the same as or similar to P02, the trouble occurrence frequency B is associated with the correlation degree w15, and the trouble occurrence frequency C is associated with the correlation degree w16. In such a case, the trouble occurrence frequency B, which has the highest correlation degree, is selected as the optimal solution. However, it is not essential to select the one with the highest correlation degree as the optimal solution, and the trouble occurrence frequency C, which has a low correlation degree but is recognized as being correlated, may be selected as the optimal solution. Furthermore, it is of course possible to select an output solution other than this that does not have an arrow connected thereto, and any other priority order may be used as long as it is based on the correlation degree.

[0127] Incidentally, when comparing the refusal rate with the reference refusal rate, if these data are expressed as average sales for a certain period, it may be determined whether they are identical or similar based on whether the average sales are within a range of ±10%. Also, if the refusal rate is shown as a time-series trend graph, it may be determined based on the similarity of the trends.

[0128] In this way, the most suitable trouble occurrence frequency can be found from the newly acquired refusal rate and displayed to the user. By looking at the search results, it is possible to determine in advance what kind of trouble occurrence frequency is likely to occur in that area in the future, and to consider the allocation of workers in each area.

[0129] FIG. 13 shows an example in which, in addition to the reference refusal rate described above, three or more levels of correlation are set between the combination with the reference population projection data and the trouble occurrence frequency for that combination.

[0130] In the example of Figure 13, the input data are, for example, reference refusal rates P01 to P03 and reference population projection data P18 to 21. The intermediate nodes shown in Figure 13 are created by combining the reference population projection data with the reference refusal rates as input data. Each intermediate node is further connected to an output. In this output, the frequency of trouble occurrence is displayed as the output solution.

[0131] Each combination (intermediate node) of the reference refusal rate and reference population estimate data is interconnected with the trouble occurrence frequency as the output solution through three or more levels of correlation. The reference refusal rate and reference population estimate data are arranged on the left side via this correlation, and the trouble occurrence frequency is arranged on the right side via this correlation. The correlation indicates the degree to which the trouble occurrence frequency is highly correlated with the reference refusal rate and reference population estimate data arranged on the left side. In other words, this correlation is an indicator of the likelihood that each reference refusal rate and reference population estimate data will be linked to what kind of trouble occurrence frequency, and indicates the accuracy of selecting the most likely trouble occurrence frequency from the reference refusal rate and reference population estimate data.

[0132] The search device 2 acquires in advance three or more levels of correlation w13 to w22 as shown in Fig. 13. In other words, when determining the actual search solution, the search device 2 accumulates past data on which was most suitable, including the reference refusal rate, the reference population estimate data obtained when acquiring the reference refusal rate, and the frequency of trouble occurrence in that case, and analyzes and interprets these to create the correlation shown in Fig. 13.

[0133] This analysis may be performed using artificial intelligence. In such a case, for example, when the reference refusal rate P01 is used and the reference population estimate data P20 is used, the trouble occurrence frequency is analyzed from past data. If there are many cases of trouble occurrence frequency A, the correlation degree connecting this trouble occurrence frequency to A is set higher, and if there are many cases of trouble occurrence frequency B and few cases of trouble occurrence frequency A, the correlation degree connecting the trouble occurrence frequency to B is set higher and the correlation degree connecting the trouble occurrence frequency to A is set lower. For example, in the example of intermediate node 61a, which is linked to the outputs of trouble occurrence frequency A and trouble occurrence frequency B, the correlation degree of w13 connecting from previous cases to trouble occurrence frequency A is set to 7 points, and the correlation degree of w14 connecting to trouble occurrence frequency B is set to 2 points.

[0134] The correlation shown in Fig. 13 may be configured by nodes of a neural network in artificial intelligence. That is, the weighting coefficients for the outputs of the nodes of this neural network correspond to the correlation described above. Furthermore, the correlation is not limited to a neural network, and may be configured by any decision-making factors that constitute artificial intelligence.

[0135] In the example of correlation degrees shown in Figure 13, node 61b is a node that combines reference refusal rate P01 with reference population estimate data P18, and the correlation degree for trouble occurrence frequency C is w15 and the correlation degree for trouble occurrence frequency E is w16. Node 61c is a node that combines reference population estimate data P19 and P21 with reference refusal rate P02, and the correlation degree for trouble occurrence frequency B is w17 and the correlation degree for trouble occurrence frequency D is w18.

[0136] This correlation is what is called "learned data" in artificial intelligence. After creating this learned data, the above-mentioned learned data will be used when actually searching for trouble occurrence frequency. In such cases, by actually entering the area to be judged for the trouble occurrence frequency, the refusal rate and population estimate data for that area will be obtained.

[0137] In this way, the optimal trouble occurrence frequency is searched for based on the newly acquired refusal rate and population projection data. In such cases, the previously acquired correlation shown in FIG. 13 (Table 1) is referenced. For example, if the newly acquired refusal rate is the same as or similar to P02 and the population projection data is the same as or similar to P21, node 61d is associated via the correlation, and this node 61d is associated with trouble occurrence frequency C at w19 and trouble occurrence frequency D at correlation w20. In such cases, the trouble occurrence frequency C with the highest correlation is selected as the optimal solution. However, it is not necessary to select the one with the highest correlation as the optimal solution; the trouble occurrence frequency D with a low correlation but a recognized correlation may be selected as the optimal solution. In addition, it is of course possible to select an output solution other than this that does not have an arrow connected thereto, and any other priority may be selected as long as it is based on the correlation.

[0138] Furthermore, reference geographical information may be used as an alternative to the above-mentioned reference information (reference population projection data, etc.).

[0139] In such a case, three or more levels of correlation between the combination of reference refusal rates and reference geographical information linked to each region in the past and the occurrence frequency of plumbing troubles are acquired in advance. Then, the refusal rates and geographical information linked to the building structure for which the occurrence frequency of plumbing troubles is predicted or the region in which it is located are acquired. Next, the occurrence frequency of plumbing troubles is searched for based on the correlation shown in Figure 13, using the reference refusal rates corresponding to the region's refusal rates in the acquired region-specific information and the reference geographical information corresponding to the acquired geographical information.

[0140] Furthermore, reference trouble type information may be used as an alternative to the above-mentioned reference information (reference population projection data, etc.).

[0141] In such a case, three or more levels of correlation between the combination of reference refusal rates and reference trouble type information linked to each region in the past and the occurrence frequency of plumbing troubles are acquired in advance. Then, the refusal rates and trouble type information linked to the building structure for which the occurrence frequency of plumbing troubles is predicted or the region in which it is located are acquired. Next, the occurrence frequency of plumbing troubles is searched for based on the correlation shown in Figure 13, using the reference refusal rates corresponding to the region's refusal rates in the acquired region-specific information and the reference trouble type information corresponding to the acquired trouble type information.

[0142] Furthermore, reference statistical information may be used as an alternative to the above-mentioned reference information (reference population projection data, etc.).

[0143] In such cases, three or more levels of correlation between the frequency of occurrence of plumbing problems and a combination of reference refusal rates linked to each region in the past and reference statistical information related to the type and age of the building structure described above is acquired in advance. Then, the refusal rate and statistical information linked to the building structure for which the frequency of occurrence of plumbing problems is predicted or the region in which it is located are acquired. Next, the frequency of occurrence of plumbing problems is searched for based on the correlation shown in Figure 13, using the reference refusal rate corresponding to the regional refusal rate in the acquired region-specific information and the reference statistical information corresponding to the acquired statistical information.

[0144] Furthermore, as an alternative to the above-mentioned reference information (reference population projection data, etc.), reference period information may be used.

[0145] In such cases, three or more levels of correlation between the combination of reference refusal rates and reference time information linked to each region in the past and the occurrence frequency of plumbing problems are acquired in advance. Then, the refusal rate and time information linked to the building structure for which the occurrence frequency of plumbing problems is to be predicted or the region in which it is located are acquired. Next, the occurrence frequency of plumbing problems is searched for based on the correlation shown in Figure 13, using the reference refusal rate corresponding to the region's refusal rate in the acquired region-specific information and the reference time information corresponding to the acquired time information.

[0146] Furthermore, reference weather information may be used instead of the above-mentioned reference information (reference population projection data, etc.).

[0147] In such cases, three or more levels of correlation between the combination of reference refusal rates and reference weather information linked to each region in the past and the occurrence frequency of plumbing problems are acquired in advance. Then, the refusal rates and weather information linked to the building structure for which the occurrence frequency of plumbing problems is predicted or the region in which it is located are acquired. Next, the occurrence frequency of plumbing problems is searched for based on the correlation shown in Figure 13, using the reference refusal rates corresponding to the region's refusal rates in the acquired region-specific information and the reference weather information corresponding to the acquired weather information.

[0148] Third embodiment The third embodiment will be described below. In carrying out this third embodiment, the trouble occurrence frequency prediction system 1, information acquisition unit 9, search device 2, and database 3 used in the first embodiment are used in the same manner. The explanation of each of these components will be omitted below by quoting the explanation of the first embodiment.

[0149] In the third embodiment, the focus is on a certain area, and the frequency of occurrence of plumbing problems in that area is predicted. In the third embodiment, reference period information and a data set of occurrence frequency of plumbing problems are learned.

[0150] 14, the input data is assumed to be reference time information P01, P02, and P03 for each region. Such reference time information P01, P02, and P03 as input data is linked to the frequency of occurrence of plumbing trouble as output.

[0151] The reference time information P01, P02, and P03 are correlated with each other through three or more levels of correlation with the plumbing trouble occurrence frequencies A to D as the output solution. The reference time information is arranged on the left side via this correlation, and the plumbing trouble occurrence frequencies are arranged on the right side via the correlation. The correlation indicates the degree to which the reference time information arranged on the left side is highly correlated with which trouble occurrence frequency. In other words, the correlation is an index that indicates which trouble occurrence frequency each piece of reference time information is likely to be linked to, and indicates the accuracy in selecting the most likely trouble occurrence frequency for each piece of reference time information. In the example of Figure 14, correlations w13 to w19 are shown. These w13 to w19 are shown on a 10-point scale as shown in Table 1 below. The closer to 10 points, the more closely each combination as an intermediate node is related to the frequency of trouble occurring as an output. Conversely, the closer to 1 point, the less closely each combination as an intermediate node is related to the frequency of trouble occurring as an output.

[0152] The search device 2 acquires in advance three or more levels of correlation w13 to w19 as shown in Fig. 14. That is, when determining an actual search solution, the search device 2 accumulates past data sets, including reference period information for each region and the frequency of trouble occurrence in that case, which was adopted and evaluated, and analyzes these to create the correlations shown in Fig. 14.

[0153] The correlations shown in Fig. 14 may be configured by nodes of a neural network in artificial intelligence. That is, the weighting coefficients for the outputs of the nodes of this neural network correspond to the correlations described above. Furthermore, the correlations are not limited to neural networks, and may be configured by any decision-making factors that constitute artificial intelligence.

[0154] This correlation is what is called "trained data" in artificial intelligence. After creating this trained data through a dataset of previous reference period information for each region and the frequency of occurrence of plumbing troubles, the trained data described above will be used to search for the frequency of troubles when actually determining the frequency of new troubles. These datasets may be created by reading them from a database managed by the contractor.

[0155] When searching for a new trouble occurrence frequency, input of time information relating to the time to be predicted is accepted.

[0156] Next, this received time information is compared with the reference time information. In such cases, the correlation degrees shown in FIG. 14 (Table 1) obtained in advance are referenced. For example, if the newly obtained time information is the same as or similar to P02, the trouble occurrence frequency B is associated with the correlation degree w15, and the trouble occurrence frequency C is associated with the correlation degree w16. In such a case, the trouble occurrence frequency B, which has the highest correlation degree, is selected as the optimal solution. However, it is not essential to select the one with the highest correlation degree as the optimal solution; the trouble occurrence frequency C, which has a low correlation degree but is recognized as being correlated, may be selected as the optimal solution. Furthermore, it is of course possible to select an output solution other than this that does not have a connecting arrow, and any other priority order may be used as long as it is based on the correlation degree.

[0157] Incidentally, when comparing the time information and the reference time information, if these data are expressed as average sales for a certain period, whether they are identical or similar may be determined based on whether the average sales is within a range of ±10%. Also, if the time information is shown as a time-series transition graph, the determination may be based on the similarity of the trends.

[0158] In this way, the most suitable trouble occurrence frequency can be searched for from newly acquired timing information and displayed to the user. By looking at the search results, it is possible to determine in advance what kind of trouble occurrence frequency is likely to occur in that area in the future, and to consider the allocation of workers in each area.

[0159] FIG. 15 shows an example in which, in addition to the above-mentioned reference period information, three or more levels of correlation are set between the combination with reference population projection data and the frequency of trouble occurrence for that combination.

[0160] In the example of Figure 15, the input data is assumed to be, for example, reference period information P01 to P03 and reference population projection data P18 to P21. The intermediate nodes shown in Figure 15 are formed by combining the reference period information as input data with the reference population projection data. Each intermediate node is further linked to an output. In this output, the frequency of trouble occurrence is displayed as an output solution.

[0161] Each combination (intermediate node) of reference period information and reference population estimate data is correlated with the trouble occurrence frequency as the output solution through three or more levels of correlation. The reference period information and reference population estimate data are arranged on the left side via this correlation, and the trouble occurrence frequency is arranged on the right side via this correlation. The correlation indicates the degree to which the trouble occurrence frequency is highly correlated with the reference period information and reference population estimate data arranged on the left side. In other words, this correlation is an index that indicates the likelihood that each reference period information and reference population estimate data will be linked to what trouble occurrence frequency, and indicates the accuracy of selecting the most likely trouble occurrence frequency from the reference period information and reference population estimate data.

[0162] The search device 2 acquires in advance three or more levels of correlation w13 to w22 as shown in Fig. 15. That is, when determining an actual search solution, the search device 2 accumulates reference period information, reference population estimate data obtained when acquiring the reference period information, and past data on which trouble occurrence frequency was most suitable in that case, and analyzes these to create the correlations shown in Fig. 15.

[0163] This analysis may be performed using artificial intelligence. In such a case, for example, when the reference period information P01 is the reference population projection data P20, the trouble occurrence frequency is analyzed from past data. If there are many cases of trouble occurrence frequency A, the degree of correlation connecting this trouble occurrence frequency to A is set higher. If there are many cases of trouble occurrence frequency B and few cases of trouble occurrence frequency A, the degree of correlation connecting the trouble occurrence frequency to B is set higher and the degree of correlation connecting the trouble occurrence frequency to A is set lower. For example, in the example of intermediate node 61a, the outputs of trouble occurrence frequency A and trouble occurrence frequency B are linked, but the degree of correlation w13 connecting from previous cases to trouble occurrence frequency A is set to 7 points, and the degree of correlation w14 connecting to trouble occurrence frequency B is set to 2 points.

[0164] The correlation shown in Fig. 15 may be configured by nodes of a neural network in artificial intelligence. That is, the weighting coefficients for the outputs of the nodes of this neural network correspond to the correlations described above. Furthermore, the correlations are not limited to neural networks, and may be configured by any decision-making factors that constitute artificial intelligence.

[0165] 15, node 61b is a node that combines reference period information P01 with reference population projection data P18, and the degree of correlation for trouble occurrence frequency C is w15 and the degree of correlation for trouble occurrence frequency E is w16. Node 61c is a node that combines reference period information P02 with reference population projection data P19 and P21, and the degree of correlation for trouble occurrence frequency B is w17 and the degree of correlation for trouble occurrence frequency D is w18.

[0166] This correlation becomes what is called "trained data" in artificial intelligence. After creating this trained data, the trained data will be used when actually searching for trouble occurrence frequency. In such cases, by actually entering the area for which the trouble occurrence frequency is to be determined, time information and population estimate data for that area will be obtained.

[0167] In this way, the optimal trouble occurrence frequency is searched for based on the newly acquired time information and population projection data. In this case, the previously acquired correlation shown in FIG. 15 (Table 1) is referenced. For example, if the newly acquired time information is the same as or similar to P02 and the population projection data is the same as or similar to P21, node 61d is associated via the correlation, and this node 61d is associated with trouble occurrence frequency C at w19 and trouble occurrence frequency D at a correlation of w20. In this case, the trouble occurrence frequency C with the highest correlation is selected as the optimal solution. However, it is not essential to select the one with the highest correlation as the optimal solution; trouble occurrence frequency D with a low correlation but a recognized correlation may also be selected as the optimal solution. Furthermore, it is of course possible to select an output solution without an arrow connecting it to the others, and any other priority order may be used as long as it is based on the correlation.

[0168] Furthermore, reference geographical information may be used as an alternative to the above-mentioned reference information (reference population projection data, etc.).

[0169] In such cases, three or more levels of correlation between the frequency of occurrence of plumbing trouble and a combination of reference time information and reference geographical information previously acquired for a certain area are acquired in advance. Then, time information related to the time to be predicted and geographical information of the area to be predicted are acquired. Next, the frequency of occurrence of plumbing trouble is searched for based on the correlation shown in Figure 15, using the reference time information corresponding to the acquired time information and the reference geographical information corresponding to the acquired geographical information.

[0170] Furthermore, reference trouble type information may be used as an alternative to the above-mentioned reference information (reference population projection data, etc.).

[0171] In such cases, three or more levels of correlation between the combination of reference time information and reference trouble type information previously acquired for a certain area and the occurrence frequency of plumbing troubles are acquired in advance. Then, time information related to the time to be predicted and trouble type information are acquired. Next, the occurrence frequency of plumbing troubles is searched for based on the correlation shown in Figure 15, using the reference time information corresponding to the acquired time information and the reference trouble type information corresponding to the acquired trouble type information.

[0172] Furthermore, reference statistical information may be used as an alternative to the above-mentioned reference information (reference population projection data, etc.).

[0173] In such cases, three or more levels of correlation between the frequency of occurrence of plumbing troubles and a combination of reference time information previously acquired for a certain area and the reference statistical information related to the type of building structure, age, etc., is acquired in advance. Then, time information related to the time to be predicted and statistical information are acquired. Next, the frequency of occurrence of plumbing troubles is searched for based on the correlation shown in Figure 15, using the reference time information corresponding to the acquired time information and the reference statistical information corresponding to the acquired statistical information.

[0174] Furthermore, reference weather information may be used instead of the above-mentioned reference information (reference population projection data, etc.).

[0175] In such cases, three or more levels of correlation between the combination of reference time information and reference weather information previously acquired for a certain area and the occurrence frequency of plumbing problems are acquired in advance. Then, the refusal rate and weather information linked to the building structure for which the occurrence frequency of plumbing problems is predicted or the area in which it is located are acquired. Next, based on the time information related to the predicted time and the reference weather information corresponding to the acquired weather information, the occurrence frequency of plumbing problems is searched for based on the correlation shown in Figure 15.

[0176] In the example of Fig. 16, input data is assumed to be reference weather information P01, P02, and P03 for each region. Such reference weather information P01, P02, and P03 as input data is linked to the frequency of occurrence of plumbing problems as output.

[0177] The reference weather information P01, P02, and P03 are correlated with each other through three or more levels of correlation to the plumbing trouble occurrence frequencies A to D as the output solution. The reference weather information is arranged on the left side via this correlation, and the plumbing trouble occurrence frequencies are arranged on the right side via this correlation. The correlation indicates the degree to which the reference weather information arranged on the left is highly correlated with which trouble occurrence frequency. In other words, this correlation is an index that indicates which trouble occurrence frequency each piece of reference weather information is likely to be linked to, and indicates the accuracy in selecting the most likely trouble occurrence frequency for each piece of reference weather information. In the example of Figure 16, correlations w13 to w19 are shown. These w13 to w19 are shown on a 10-point scale as shown in Table 1 below. The closer to 10 points, the more closely each combination as an intermediate node is related to the frequency of trouble occurring as an output. Conversely, the closer to 1 point, the less closely each combination as an intermediate node is related to the frequency of trouble occurring as an output.

[0178] The search device 2 acquires in advance three or more levels of correlation w13 to w19 as shown in Fig. 16. That is, when determining an actual search solution, the search device 2 accumulates past data sets, including the reference weather information for each region and the frequency of trouble occurrence in that case, which was adopted and evaluated, and analyzes these to create the correlations shown in Fig. 16.

[0179] The correlations shown in Fig. 16 may be configured by nodes of a neural network in artificial intelligence. That is, the weighting coefficients for the outputs of the nodes of this neural network correspond to the correlations described above. Furthermore, the correlations are not limited to neural networks, and may be configured by any decision-making factors that constitute artificial intelligence.

[0180] This correlation is what is called "trained data" in artificial intelligence. After creating this trained data using a dataset of previous reference weather information for each region and the frequency of occurrence of plumbing problems, the trained data will be used to search for the frequency of occurrence of new problems when actually determining the frequency of occurrence of problems. These datasets may be created by reading them from a database managed by the contractor.

[0181] When searching for a new frequency of trouble occurrence, weather information for the time period to be predicted is input.

[0182] Next, the received weather information is compared with the reference weather information. In such cases, the correlation degrees shown in FIG. 16 (Table 1) obtained in advance are referenced. For example, if the newly obtained weather information is the same as or similar to P02, the trouble occurrence frequency B is associated with a correlation degree of w15, and the trouble occurrence frequency C is associated with a correlation degree of w16. In such a case, the trouble occurrence frequency B, which has the highest correlation degree, is selected as the optimal solution. However, it is not essential to select the one with the highest correlation degree as the optimal solution; the trouble occurrence frequency C, which has a low correlation degree but is recognized as being correlated, may also be selected as the optimal solution. Of course, it is also possible to select an output solution other than this that has no arrows connected to it, and any other priority order may be used as long as it is based on the correlation degree.

[0183] Incidentally, when comparing the weather information and the reference weather information, if these data are expressed as average sales for a certain period, whether they are identical or similar may be determined based on whether the average sales are within a range of ±10%. Also, if the weather information is shown as a time-series trend graph, the determination may be based on the similarity of the trends.

[0184] In this way, the most suitable trouble occurrence frequency can be searched for from newly acquired weather information and displayed to the user. By looking at the search results, it is possible to determine in advance what kind of trouble occurrence frequency is likely to occur in that area in the future, and to consider the allocation of workers in each area.

[0185] FIG. 17 shows an example in which three or more levels of correlation are set between the reference weather information described above, the reference population projection data, and the trouble occurrence frequency for that combination.

[0186] In the example of Figure 17, the input data is assumed to be, for example, reference weather information P01 to P03 and reference population projection data P18 to P21. The intermediate nodes shown in Figure 17 are formed by combining the reference weather information as input data with the reference population projection data. Each intermediate node is further connected to an output. In this output, the frequency of trouble occurrence is displayed as an output solution.

[0187] Each combination (intermediate node) of reference weather information and reference population estimate data is correlated with the trouble occurrence frequency as the output solution through three or more levels of correlation. The reference weather information and reference population estimate data are arranged on the left side via this correlation, and the trouble occurrence frequency is arranged on the right side via this correlation. The correlation indicates the degree to which the trouble occurrence frequency is highly correlated with the reference weather information and reference population estimate data arranged on the left side. In other words, this correlation is an index that indicates the likelihood that each piece of reference weather information and reference population estimate data will be linked to what kind of trouble occurrence frequency, and indicates the accuracy of selecting the most likely trouble occurrence frequency from the reference weather information and reference population estimate data.

[0188] The search device 2 acquires in advance three or more levels of correlation w13 to w22 as shown in Fig. 17. That is, when determining an actual search solution, the search device 2 accumulates reference weather information, reference population estimate data obtained when acquiring the reference weather information, and past data on which of the trouble occurrence frequencies was most suitable, and analyzes these to create the correlations shown in Fig. 15.

[0189] This analysis may be performed using artificial intelligence. In such a case, for example, when reference weather information P01 and reference population estimate data P20 are used, the trouble occurrence frequency is analyzed from past data. If there are many cases of trouble occurrence frequency A, the correlation degree connecting this trouble occurrence frequency to A is set higher. If there are many cases of trouble occurrence frequency B and few cases of trouble occurrence frequency A, the correlation degree connecting the trouble occurrence frequency to B is set higher and the correlation degree connecting the trouble occurrence frequency to A is set lower. For example, in the example of intermediate node 61a, the outputs of trouble occurrence frequency A and trouble occurrence frequency B are linked, but the correlation degree of w13 connecting from previous cases to trouble occurrence frequency A is set to 7 points, and the correlation degree of w14 connecting to trouble occurrence frequency B is set to 2 points.

[0190] The correlations shown in Fig. 17 may be configured by nodes of a neural network in artificial intelligence. That is, the weighting coefficients for the outputs of the nodes of this neural network correspond to the correlations described above. Furthermore, the correlations are not limited to neural networks, and may be configured by any decision-making factors that constitute artificial intelligence.

[0191] 17, node 61b is a node that combines reference weather information P01 with reference population estimate data P18, and the degree of correlation for trouble occurrence frequency C is w15 and the degree of correlation for trouble occurrence frequency E is w16. Node 61c is a node that combines reference weather information P02 with reference population estimate data P19 and P21, and the degree of correlation for trouble occurrence frequency B is w17 and the degree of correlation for trouble occurrence frequency D is w18.

[0192] This correlation is what is called "trained data" in artificial intelligence. After creating this trained data, the trained data will be used when searching for the frequency of trouble occurrence. In such cases, by actually entering the area for which the frequency of trouble occurrence is to be determined, weather information and population estimate data for that area will be obtained.

[0193] In this way, the optimal trouble occurrence frequency is searched for based on the newly acquired weather information and population estimate data. In this case, the previously acquired correlation shown in FIG. 17 (Table 1) is referenced. For example, if the newly acquired weather information is identical to or similar to P02 and the population estimate data is identical to or similar to P21, node 61d is associated via the correlation, and this node 61d is associated with trouble occurrence frequency C at w19 and trouble occurrence frequency D at a correlation of w20. In this case, the trouble occurrence frequency C with the highest correlation is selected as the optimal solution. However, it is not essential to select the one with the highest correlation as the optimal solution; trouble occurrence frequency D with a low correlation but a recognized correlation may be selected as the optimal solution. Of course, other output solutions without connected arrows may also be selected, and any other priority order may be used as long as it is based on the correlation.

[0194] Furthermore, reference geographical information may be used as an alternative to the above-mentioned reference information (reference population projection data, etc.).

[0195] In such cases, three or more levels of correlation between a combination of reference weather information and reference geographical information previously acquired for a certain area and the occurrence frequency of plumbing troubles are acquired in advance. Then, weather information related to the time period to be predicted and geographical information for the area to be predicted are acquired. Next, the occurrence frequency of plumbing troubles is searched for based on the correlation shown in Figure 17, using the reference weather information corresponding to the acquired weather information and the reference geographical information corresponding to the acquired geographical information.

[0196] Furthermore, reference trouble type information may be used as an alternative to the above-mentioned reference information (reference population projection data, etc.).

[0197] In such cases, three or more levels of correlation between a combination of reference weather information and reference trouble type information previously acquired for a certain area and the occurrence frequency of plumbing troubles are acquired in advance. Then, weather information and trouble type information related to the time period to be predicted are acquired. Next, the occurrence frequency of plumbing troubles is searched for based on the correlation shown in Figure 17, using the reference weather information corresponding to the acquired weather information and the reference trouble type information corresponding to the acquired trouble type information.

[0198] Furthermore, reference statistical information may be used as an alternative to the above-mentioned reference information (reference population projection data, etc.).

[0199] In such cases, three or more levels of correlation between the frequency of occurrence of plumbing trouble and a combination of reference weather information previously acquired for a certain area and the reference statistical information related to the type of building structure, age, etc., is acquired in advance. Then, weather information and statistical information related to the time to be predicted are acquired. Next, the frequency of occurrence of plumbing trouble is searched for based on the correlation shown in Figure 17, using the reference weather information corresponding to the acquired weather information and the reference statistical information corresponding to the acquired statistical information.

[0200] Fourth embodiment The fourth embodiment will be described below. In carrying out this fourth embodiment, the trouble occurrence frequency prediction system 1, information acquisition unit 9, search device 2, and database 3 used in the first embodiment are used in the same manner. The explanation of each of these components will be omitted below by quoting the explanation of the first embodiment.

[0201] In the fourth embodiment, reference geographical information and a dataset of occurrence frequencies of plumbing problems are learned.

[0202] In the example of Fig. 18, the input data is geographical information references P01, P02, and P03 for each region. The geographical information references P01, P02, and P03 as input data are linked to the frequency of occurrence of plumbing problems as output.

[0203] The reference geographical information P01, P02, and P03 are correlated with each other through three or more levels of correlation with the plumbing trouble occurrence frequencies A to D as the output solution. The reference geographical information is arranged on the left side via this correlation, and the plumbing trouble occurrence frequencies are arranged on the right side via this correlation. The correlation indicates the degree to which the reference geographical information arranged on the left side is highly correlated with which trouble occurrence frequency. In other words, the correlation is an index that indicates the likelihood that each piece of reference geographical information is linked to which trouble occurrence frequency, and indicates the accuracy in selecting the most likely trouble occurrence frequency for each piece of reference geographical information. In the example of Figure 18, correlations w13 to w19 are shown. These w13 to w19 are shown on a 10-point scale as shown in Table 1 below. The closer to 10 points, the more closely each combination as an intermediate node is related to the frequency of trouble occurring as an output. Conversely, the closer to 1 point, the less closely each combination as an intermediate node is related to the frequency of trouble occurring as an output.

[0204] The search device 2 acquires in advance three or more levels of correlation w13 to w19 as shown in Fig. 18. That is, when determining an actual search solution, the search device 2 accumulates past data sets, including reference geographical information for each region and the frequency of trouble occurrence in that case, which was adopted and evaluated, and analyzes these to create the correlations shown in Fig. 18.

[0205] The correlation shown in Fig. 18 may be configured by nodes of a neural network in artificial intelligence. That is, the weighting coefficients for the outputs of the nodes of this neural network correspond to the correlations described above. Furthermore, the correlations are not limited to neural networks, and may be configured by any decision-making factors that constitute artificial intelligence.

[0206] This correlation is what is called "trained data" in artificial intelligence. After creating this trained data through a dataset of previous reference geographical information for each region and the frequency of occurrence of plumbing problems, the trained data will be used to search for the frequency of occurrence of problems when actually determining the frequency of new problems. These datasets may be created by reading them from a database managed by the contractor.

[0207] When searching for a new frequency of trouble occurrence, input of geographical information regarding the time period to be predicted is accepted.

[0208] Next, this received geographical information is compared with the reference geographical information. In such cases, the correlation degrees shown in FIG. 18 (Table 1) obtained in advance are referenced. For example, if the newly obtained geographical information is identical to or similar to P02, the trouble occurrence frequency B is associated with the correlation degree w15, and the trouble occurrence frequency C is associated with the correlation degree w16. In such a case, the trouble occurrence frequency B, which has the highest correlation degree, is selected as the optimal solution. However, it is not essential to select the one with the highest correlation degree as the optimal solution; the trouble occurrence frequency C, which has a low correlation degree but is recognized as being correlated, may be selected as the optimal solution. Furthermore, it is of course possible to select an output solution other than this that does not have a connecting arrow, and any other priority order may be used as long as it is based on the correlation degree.

[0209] Incidentally, when comparing geographical information with reference geographical information, if these data are expressed as average sales for a certain period, whether they are identical or similar may be determined based on whether the average sales is within a range of ±10%. Also, if the geographical information is shown as a time-series trend graph, the determination may be based on the similarity of the trends.

[0210] In this way, the most suitable trouble occurrence frequency can be searched for from newly acquired geographical information and displayed to the user. By looking at the search results, it is possible to determine in advance what kind of trouble occurrence frequency is likely to occur in that area in the future, and to consider the allocation of workers in each area.

[0211] FIG. 19 shows an example in which three or more levels of correlation are set between the reference geographical information described above, the reference population projection data, and the frequency of trouble occurrence for that combination.

[0212] In the example of Figure 19, the input data is assumed to be, for example, reference geographical information P01 to P03 and reference population projection data P18 to P21. The intermediate nodes shown in Figure 19 are formed by combining the reference geographical information as input data with the reference population projection data. Each intermediate node is further connected to an output. In this output, the frequency of trouble occurrence is displayed as an output solution.

[0213] Each combination (intermediate node) of reference geographic information and reference population estimate data is related to the trouble occurrence frequency as the output solution through three or more levels of correlation. The reference geographic information and reference population estimate data are arranged on the left side via this correlation, and the trouble occurrence frequency is arranged on the right side via this correlation. The correlation indicates the degree to which the trouble occurrence frequency is highly related to the reference geographic information and reference population estimate data arranged on the left side. In other words, this correlation is an index that indicates the likelihood that each reference geographic information and reference population estimate data will be linked to what trouble occurrence frequency, and indicates the accuracy of selecting the most likely trouble occurrence frequency from the reference geographic information and reference population estimate data.

[0214] The search device 2 acquires in advance three or more levels of correlation w13 to w22 as shown in Fig. 19. That is, when determining an actual search solution, the search device 2 accumulates reference geographical information, reference population estimate data obtained when acquiring the reference geographical information, and past data on which of the trouble occurrence frequencies was most suitable, and analyzes these to create the correlations shown in Fig. 19.

[0215] This analysis may be performed using artificial intelligence. In such a case, for example, when the reference geographic information P01 is the reference population projection data P20, the trouble occurrence frequency is analyzed from past data. If there are many cases of trouble occurrence frequency A, the correlation degree connecting this trouble occurrence frequency to A is set higher. If there are many cases of trouble occurrence frequency B and few cases of trouble occurrence frequency A, the correlation degree connecting the trouble occurrence frequency to B is set higher and the correlation degree connecting the trouble occurrence frequency to A is set lower. For example, in the example of intermediate node 61a, the outputs of trouble occurrence frequency A and trouble occurrence frequency B are linked, but the correlation degree of w13 connecting from previous cases to trouble occurrence frequency A is set to 7 points, and the correlation degree of w14 connecting to trouble occurrence frequency B is set to 2 points.

[0216] 19 may be configured by nodes of a neural network in artificial intelligence. That is, the weighting coefficients for the outputs of the nodes of this neural network correspond to the above-mentioned degrees of association. Furthermore, the degrees of association are not limited to neural networks, and may be configured by any decision-making factors that constitute artificial intelligence.

[0217] 19, node 61b is a node that represents a combination of reference geographical information P01 and reference population projection data P18, and the degree of correlation for trouble occurrence frequency C is w15 and the degree of correlation for trouble occurrence frequency E is w16. Node 61c is a node that represents a combination of reference population projection data P19 and P21 for reference geographical information P02, and the degree of correlation for trouble occurrence frequency B is w17 and the degree of correlation for trouble occurrence frequency D is w18.

[0218] This correlation is what is called "trained data" in artificial intelligence. After creating this trained data, the trained data will be used when searching for the frequency of trouble occurrence. In such cases, by actually entering the area for which the frequency of trouble occurrence is to be determined, geographical information and population estimate data for that area will be obtained.

[0219] In this way, the optimal trouble occurrence frequency is searched for based on the newly acquired geographical information and population projection data. In this case, the previously acquired correlation shown in FIG. 19 (Table 1) is referenced. For example, if the newly acquired geographical information is identical to or similar to P02 and the population projection data is identical to or similar to P21, node 61d is associated via the correlation, and this node 61d is associated with trouble occurrence frequency C at w19 and trouble occurrence frequency D at a correlation of w20. In this case, the trouble occurrence frequency C with the highest correlation is selected as the optimal solution. However, it is not essential to select the one with the highest correlation as the optimal solution; trouble occurrence frequency D with a low correlation but a recognized correlation may be selected as the optimal solution. Furthermore, it is of course possible to select an output solution without an arrow connecting it to the others, and any other priority order may be used as long as it is based on the correlation.

[0220] Furthermore, reference trouble type information may be used as an alternative to the above-mentioned reference information (reference population projection data, etc.).

[0221] In such cases, three or more levels of correlation between the combination of reference geographical information and reference trouble type information previously acquired for a certain area and the occurrence frequency of plumbing troubles are acquired in advance. Then, geographical information and trouble type information for the area to be predicted are acquired. Next, the occurrence frequency of plumbing troubles is searched for based on the correlation shown in Figure 15, using the reference geographical information corresponding to the acquired geographical information and the reference trouble type information corresponding to the acquired trouble type information.

[0222] Furthermore, reference statistical information may be used as an alternative to the above-mentioned reference information (reference population projection data, etc.).

[0223] In such cases, three or more levels of correlation between a combination of reference geographical information previously acquired for a certain area and the reference statistical information related to the type of building structure, age, etc., and the occurrence frequency of plumbing problems are acquired in advance. Then, geographical information and statistical information related to the area to be predicted are acquired. Next, the occurrence frequency of plumbing problems is searched for based on the correlation shown in Figure 15, using the reference geographical information corresponding to the acquired geographical information and the reference statistical information corresponding to the acquired statistical information.

[0224] In the example of Fig. 20, the input data are reference statistical information P01, P02, and P03 regarding the types of building structures in each region. Such reference statistical information P01, P02, and P03 as input data are linked to the frequency of plumbing troubles as output.

[0225] The reference statistical information P01, P02, and P03 are correlated with each other through three or more levels of correlation with the plumbing trouble occurrence frequencies A to D as the output solution. The reference statistical information is arranged on the left side via this correlation, and the plumbing trouble occurrence frequencies are arranged on the right side via the correlation. The correlation indicates the degree to which the reference statistical information arranged on the left side is highly correlated with the trouble occurrence frequency. In other words, the correlation is an index that indicates which trouble occurrence frequency each reference statistical information is likely to be linked to, and indicates the accuracy in selecting the most likely trouble occurrence frequency for each reference statistical information. In the example of Figure 20, correlations w13 to w19 are shown. These w13 to w19 are shown on a 10-point scale as shown in Table 1 below. The closer to 10 points, the more closely each combination as an intermediate node is related to the frequency of trouble occurring as an output. Conversely, the closer to 1 point, the less closely each combination as an intermediate node is related to the frequency of trouble occurring as an output.

[0226] The search device 2 acquires in advance three or more levels of correlation w13 to w19 as shown in Fig. 20. That is, when determining an actual search solution, the search device 2 accumulates past data sets, including reference statistical information for each region and the frequency of trouble occurrence in that case, which was adopted and evaluated, and analyzes these to create the correlations shown in Fig. 20.

[0227] The correlation shown in Fig. 20 may be configured by nodes of a neural network in artificial intelligence. That is, the weighting coefficients for the outputs of the nodes of this neural network correspond to the correlations described above. Furthermore, the correlations are not limited to neural networks, and may be configured by any decision-making factors that constitute artificial intelligence.

[0228] This correlation is what is called "trained data" in artificial intelligence. After creating this trained data through a dataset of previous reference statistical information for each region and the frequency of occurrence of plumbing problems, the trained data described above will be used to search for the frequency of occurrence of problems when actually determining the frequency of new problems. These datasets may be created by reading them from a database managed by the contractor.

[0229] When searching for a new frequency of trouble occurrence, statistical information for the area to be predicted is accepted as input.

[0230] Next, this received statistical information is compared with the reference statistical information. In such cases, the correlation degrees shown in FIG. 20 (Table 1) obtained in advance are referenced. For example, if the newly obtained statistical information is identical to or similar to P02, the trouble occurrence frequency B is associated with the correlation degree w15, and the trouble occurrence frequency C is associated with the correlation degree w16. In such a case, the trouble occurrence frequency B, which has the highest correlation degree, is selected as the optimal solution. However, it is not essential to select the one with the highest correlation degree as the optimal solution; the trouble occurrence frequency C, which has a low correlation degree but is recognized as being correlated, may be selected as the optimal solution. Furthermore, it is of course possible to select an output solution other than this that has no arrows connected to it, and any other priority order may be used as long as it is based on the correlation degree.

[0231] Incidentally, when comparing statistical information and reference statistical information, if these data are expressed as average sales for a certain period, whether they are identical or similar may be determined based on whether the average sales is within a range of ±10%. Also, if the statistical information is shown as a time-series trend graph, the determination may be based on the similarity of the trends.

[0232] In this way, the most suitable trouble occurrence frequency can be found from newly acquired statistical information and displayed to the user. By looking at the search results, it is possible to determine in advance what kind of trouble occurrence frequency is likely to occur in that area in the future, and to consider the allocation of workers in each area. The reference statistical information and statistical information are not limited to statistics relating to the type of building structure, but may also reflect various statistics such as the age of the building.

[0233] FIG. 21 shows an example in which, in addition to the above-mentioned reference statistical information, a combination with reference population projection data is set, and three or more levels of correlation are set between the frequency of trouble occurrence for that combination.

[0234] In the example of Figure 21, the input data is assumed to be, for example, reference statistical information P01 to P03 and reference population projection data P18 to P21. The intermediate nodes shown in Figure 21 are formed by combining the reference statistical information as input data with the reference population projection data. Each intermediate node is further linked to an output. In this output, the frequency of trouble occurrence is displayed as an output solution.

[0235] Each combination (intermediate node) of reference statistical information and reference population estimate data is interconnected through three or more levels of correlation with respect to the trouble occurrence frequency as the output solution. The reference statistical information and reference population estimate data are arranged on the left side via this correlation, and the trouble occurrence frequency is arranged on the right side via this correlation. The correlation indicates the degree to which the trouble occurrence frequency is highly related to the reference statistical information and reference population estimate data arranged on the left side. In other words, this correlation is an index that indicates the likelihood that each piece of reference statistical information and reference population estimate data will be linked to what kind of trouble occurrence frequency, and indicates the accuracy of selecting the most likely trouble occurrence frequency from the reference statistical information and reference population estimate data.

[0236] The search device 2 acquires in advance three or more levels of correlation w13 to w22 as shown in Fig. 21. That is, when determining an actual search solution, the search device 2 accumulates reference statistical information, reference population projection data obtained when acquiring the reference statistical information, and past data on which trouble occurrence frequency was most suitable in that case, and analyzes these to create the correlations shown in Fig. 21.

[0237] This analysis may be performed using artificial intelligence. In such a case, for example, when the reference statistical information P01 is the reference population projection data P20, the trouble occurrence frequency is analyzed from past data. If there are many cases of trouble occurrence frequency A, the degree of correlation connecting this trouble occurrence frequency to A is set higher. If there are many cases of trouble occurrence frequency B and few cases of trouble occurrence frequency A, the degree of correlation connecting the trouble occurrence frequency to B is set higher and the degree of correlation connecting the trouble occurrence frequency to A is set lower. For example, in the example of intermediate node 61a, the outputs of trouble occurrence frequency A and trouble occurrence frequency B are linked, but the degree of correlation w13 connecting from previous cases to trouble occurrence frequency A is set to 7 points, and the degree of correlation w14 connecting to trouble occurrence frequency B is set to 2 points.

[0238] The correlations shown in Fig. 21 may be configured by nodes of a neural network in artificial intelligence. That is, the weighting coefficients for the outputs of the nodes of this neural network correspond to the correlations described above. Furthermore, the correlations are not limited to neural networks, and may be configured by any decision-making factors that constitute artificial intelligence.

[0239] In the example of correlation degrees shown in Figure 21, node 61b is a node that combines reference statistical information P01 with reference population projection data P18, and the correlation degree for trouble occurrence frequency C is w15 and the correlation degree for trouble occurrence frequency E is w16. Node 61c is a node that combines reference population projection data P19 and P21 with reference statistical information P02, and the correlation degree for trouble occurrence frequency B is w17 and the correlation degree for trouble occurrence frequency D is w18.

[0240] This correlation becomes what is called "trained data" in artificial intelligence. After creating this trained data, the trained data will be used when searching for the frequency of trouble occurrence. In such cases, by actually entering the area for which the frequency of trouble occurrence is to be determined, statistical information and population estimate data for that area will be obtained.

[0241] In this way, the optimal trouble occurrence frequency is searched for based on the newly acquired statistical information and population projection data. In this case, the previously acquired correlation shown in FIG. 21 (Table 1) is referenced. For example, if the newly acquired statistical information is identical to or similar to P02 and the population projection data is identical to or similar to P21, node 61d is associated via the correlation, and this node 61d is associated with trouble occurrence frequency C at w19 and trouble occurrence frequency D at a correlation of w20. In this case, the trouble occurrence frequency C with the highest correlation is selected as the optimal solution. However, it is not essential to select the one with the highest correlation as the optimal solution; trouble occurrence frequency D with a low correlation but in which correlation is recognized may be selected as the optimal solution. Of course, other output solutions without connected arrows may also be selected, and any other priority may be selected based on the correlation.

[0242] Furthermore, reference trouble type information may be used as an alternative to the above-mentioned reference information (reference population projection data, etc.).

[0243] In such cases, three or more levels of correlation between the frequency of occurrence of plumbing troubles and a combination of reference statistical information and reference trouble type information previously acquired for a certain area are acquired in advance. Then, statistical information related to the time period to be predicted and trouble type information are acquired. Next, the frequency of occurrence of plumbing troubles is searched for based on the correlation shown in Figure 21, using the reference statistical information corresponding to the acquired statistical information and the reference trouble type information corresponding to the acquired trouble type information.

[0244] Note that, in this embodiment, the reference geographical information and the reference statistical information are used as a base, and other reference information (reference population estimate data, reference trouble type information, etc.) is combined with this to obtain in advance three or more levels of correlation with the occurrence frequency of plumbing troubles, but the present invention is not limited to this. Of course, any of the reference information in the first to fourth embodiments may be used as a base, and by combining it with other reference information, three or more levels of correlation with the occurrence frequency of plumbing troubles may be obtained in advance, and a solution search may be performed.

[0245] The present invention is not limited to the above-described embodiment, and may utilize, for example, three or more levels of correlation between the basic reference information and the occurrence frequency of plumbing troubles, as shown in Fig. 22. In such a case, a solution search is performed based on three or more levels of correlation with the occurrence frequency of plumbing troubles according to newly acquired information. For example, all of the reference information from the first embodiment onward can be applied as the basic reference information.

[0246] Similarly, in these cases, when information corresponding to the reference information used as learning data is input, a solution search is performed based on the above-described method.

[0247] The search solution determined through the association may further be modified or weighted based on other reference information.

[0248] The other reference information referred to here corresponds to any reference information other than the reference information that is the base reference information, when any of the above-mentioned reference information is used as the base reference information.

[0249] For example, suppose that in one piece of other reference information, reference information P02, the occurrence frequency of a plumbing problem has often been determined to be B. When a new occurrence frequency of a plumbing problem according to this reference information P02 is obtained, a process is performed to increase the weight of search solution B as the occurrence frequency of the plumbing problem, in other words, a process is set in advance to lead to search solution B as the occurrence frequency of the plumbing problem.

[0250] For example, let us assume that other reference information G is an analysis result that more strongly suggests search solution C as the occurrence frequency of plumbing troubles, and reference information F is an analysis result that more strongly suggests search solution D as the occurrence frequency of plumbing troubles. After this setting between the reference information, if the actually acquired information is identical or similar to reference information G, a process is performed to increase the weighting of search solution C as the occurrence frequency of plumbing troubles. Conversely, if the actually acquired information is identical or similar to reference information F, a process is performed to increase the weighting of search solution D as the occurrence frequency of plumbing troubles. In other words, the correlation itself leading to the occurrence frequency of plumbing troubles may be controlled based on this reference information F to H. Alternatively, the occurrence frequency of plumbing troubles may be determined based only on the above-mentioned correlation, and then the obtained search solution may be modified based on reference information F to H. In the latter case, the weighting and the degree of modification of the occurrence frequency of plumbing troubles as the search solution based on reference information F to H will be reflected in each case by the system design.

[0251] Furthermore, the reference information is not limited to being composed of one type, and a solution search may be performed based on two or more types of reference information. Similarly, in such a case, the more a case is linked to the occurrence frequency of a plumbing problem suggested by the reference information, the higher the occurrence frequency of the plumbing problem as a search solution obtained through the correlation may be revised.

[0252] Similarly, as shown in Fig. 23, when forming a correlation between a combination of basic reference information and other reference information and the occurrence frequency of a plumbing problem, any of the reference information from the first embodiment onward can be applied as the basic reference information. The other reference information includes any reference information other than the basic reference information.

[0253] In this case, if the basic reference information is reference statistical information, the other reference information includes any other reference information.

[0254] In such cases, the occurrence frequency of the plumbing trouble can be estimated by performing a solution search in the same way. In this case, as shown in Fig. 22 above, the occurrence frequency of the plumbing trouble may be corrected for the search solution obtained through the correlation degree using further reference information (reference information F, G, H, etc.).

[0255] As shown in FIGS. 24 and 25, a solution search may be performed using three or more levels of correlation between basic reference information and the occurrence frequency of plumbing troubles.

[0256] The transaction price is determined based on the reference information only. For example, as shown in Figures 24 and 25, the correlation between the reference information acquired in the past (including any reference information from the first embodiment onward) and the frequency of occurrence of plumbing troubles actually determined in the past is used, with three or more levels of correlation.

[0257] This correlation is what is called "trained data" in artificial intelligence. After creating this trained data, the trained data will be used when determining new transaction prices. In such cases, new information corresponding to the reference information will be acquired.

[0258] Based on the newly acquired information in this way, the frequency of occurrence of plumbing troubles is determined. In such cases, the correlation degrees shown in Figures 24 and 25, which have been acquired in advance, are referenced. The specific estimation method is the same as that described above, so a detailed explanation will be omitted below.

[0259] Fifth embodiment In the fifth embodiment, instead of searching for the frequency of occurrence of plumbing troubles, the repair costs, repair periods, and repair methods for plumbing troubles are searched for.

[0260] Repair costs include all expenses incurred to fix plumbing problems. These expenses include toilet clogs, leaks, and breakdowns, kitchen and bathroom clogs, leaks, and breakdowns, as well as all expenses incurred to restore plumbing problems to normal, such as clogged toilets, leaks, and breakdowns in the kitchen and bathroom, clogged bath drains, leaks in showers and faucets, and breakdowns. Repairs here include not only actual plumbing problems, but also preventative repairs and maintenance for things that are not currently causing problems but could lead to problems if left unattended. Repairs also include the removal of dirt that is not damaged but has accumulated. Repairs also include the removal of odors and other defects that have existed since purchase, even if they are not damaged.

[0261] The repair period refers to the period from the date of repair or application for repair in the case of the above-mentioned plumbing trouble until the repair is completed. The repair method refers to the specific repair content for the above-mentioned damage, and includes methods such as complete replacement, partial replacement, replacement with a compatible part, retightening bolts, and even methods for unclogging toilets, leaking water, clogged kitchens and bathrooms, clogged bathtub drains, and leaking showers and faucets.

[0262] A detailed explanation of the solution search when the search solution is either repair costs, repair period, or repair method will be given by replacing the frequency of occurrence of plumbing problems from the first embodiment onwards with repair costs, repair period, and repair method, and will not be repeated below.

[0263] Specifically, by similarly learning the learning data sets of repair costs, repair periods, and repair methods for each of the above-mentioned reference information, it becomes possible to similarly perform a solution search. The reference information applicable in this fifth embodiment is mainly reference trouble type information. The reference trouble type information or trouble type information referred to here may be configured not only by the type of trouble as described above, but also by the degree of the trouble. In other words, in the case of a clogged toilet, for example, the reference trouble type information may be a score indicating the degree of clogged toilet or may indicate the severity of the clog.

[0264] Such reference trouble type information and trouble type information may be obtained by capturing images of the target of the plumbing trouble (toilet, bath, etc.). The images referred to here may be not only still images but also moving images. For example, in the case of a clogged toilet, the extent of the clog can often be determined by capturing a moving image of the state inside the toilet bowl after flushing. In such cases, the captured images may be analyzed, and the feature amounts, etc. may be analyzed using deep learning technology, etc., to create a pattern for the extent of the plumbing trouble.

[0265] Reference contract information may be applied as reference information applicable in this fifth embodiment. Reference contract information is all information related to the contract contents of repair support for plumbing problems. When purchasing a new building, toilet, bath, or kitchen, various warranty and repair support contracts are often concluded, and such repair support often clearly specifies the extent of repair support and the type of damage. The reference contract information represents the specific repair support content for the plumbing problem or damage conditions, as well as the costs and ratios to be borne by the contractor and the customer, respectively. Specifically, repair support conditions (period, cost, ratio), etc. may be extracted and trained. When training such reference contract information, contract information is acquired when a new solution search is performed. The data type of this contract information is the same as that of the reference contract information.

[0266] Reference odor information may be used as reference information applicable to this fifth embodiment. This reference odor information is information obtained by sensing odors and odors around wet areas. This reference odor information can be detected, for example, by an odor sensor. To detect reducing odors such as hydrogen sulfide, acetaldehyde, and ammonia, the reference odor information may be composed of a semiconductor-type odor sensor that utilizes the adsorption of odor molecules on a conductor surface and the change in semiconductor resistance due to the surface reaction, a quartz crystal oscillator-type odor sensor with an odor-sensing membrane made of a lipid membrane made of natural or synthetic lipids that selectively adsorb molecules attached to the surface of a quartz crystal oscillator, or a selective FET biosensor that detects molecules in the air. When learning such reference odor information, odor information is acquired when a new solution search is performed. The type of data for this odor information is the same as that of the reference odor information.

[0267] Reference sound information may be used as reference information applicable in this fifth embodiment. The reference sound information is a recording of the sounds of the plumbing area to be distinguished. That is, this reference sound information includes the sound of flushing a toilet or running water in the kitchen, abnormal sound volume, resonance, reverberation, abnormal sounds, etc. The sound data constituting this reference sound information may be composed of sound wave data on the time axis, for example, or may be converted to the frequency axis to indicate the frequency band of the sound being generated. When learning such reference sound information, sound information is acquired when a new solution search is performed. The type of data for this sound information is the same as that of the reference sound information.

[0268] In this fifth embodiment, reference location information or location information may be used as applicable reference information. The reference location information or location information here is information about the location where the plumbing problem occurred, and may be indicated by an address, GPS coordinates, or the like. When actually calculating the repair cost and repair period, it is necessary to take into account the address and location of the house where the plumbing problem occurred, so this is added to the explanatory variables.

[0269] In the fifth embodiment, the quality of the tray, bath, etc. that make up the wet area may be searched for, and the repair cost, repair period, and repair method may be searched for based on the searched quality. As mentioned above, the quality here refers to an evaluation of the state of the wet area from various perspectives such as beauty, wear, feeling of use, smell, scratches, noise, and damage, but in addition to this, in the fifth embodiment, the degree of damage if there is damage, the details of the damage, and the degree of discomfort or malfunction if there is any.

[0270] The solution search method for the quality of plumbing will be omitted from the following description by replacing the solution search in the first to fourth embodiments with the quality of plumbing.

[0271] Based on the quality of the plumbing obtained in this way, repair costs, repair periods, and repair methods are calculated. In such cases, a template is prepared in advance in which repair costs, repair periods, and repair methods are linked to each quality of the plumbing. In other words, once the quality of the plumbing is determined, the corresponding repair costs, repair periods, and repair methods can be easily calculated by referencing the template. This makes it possible to determine the quality of the plumbing from various information, and from that, estimate the repair costs, repair periods, and repair methods.

[0272] Similarly, the insurance premium to be paid to the contractor or customer may be calculated from the quality of the plumbing. In such a case, a template is prepared in advance in which insurance premiums are linked to each quality of the plumbing. Once the quality of the building is determined, the corresponding insurance premium can be easily calculated by referring to the template. This makes it possible to determine the quality of the building from various information and estimate the insurance premium from that.

[0273] When calculating the insurance premium, the repair cost, repair period, and repair method may be searched and the insurance premium may be calculated from these. In such a case, a template is prepared in advance in which the insurance premium is linked to each repair cost, repair period, and repair method. Once the repair cost, repair period, and repair method are determined, the corresponding insurance premium can be easily calculated by referring to the template. This makes it possible to determine the repair cost, repair period, and repair method from various information and estimate the insurance premium from this.

[0274] Of course, in the fifth embodiment as well, solution searches may be performed as shown in Figures 22 to 25. Furthermore, the targets for searching repair costs, repair periods, and repair methods may be buildings, machines, plumbing (washrooms, baths, kitchens, toilets), equipment, and home appliances in addition to buildings. [Explanation of symbols]

[0275] 1. Plumbing trouble occurrence frequency system 2 Search device 21 Internal Bus 23 Display section 24 Control Unit 25 Control section 26 Communications Department 27 Discrimination part 28 Memory section 61 nodes

Claims

1. In a plumbing trouble repair cost estimation program that estimates repair costs for plumbing troubles, an information acquisition step of acquiring trouble type information relating to the type of plumbing trouble and contract information relating to the contract details of the plumbing trouble repair support; and an estimation step of estimating repair costs by using a trained model in which a combination of previously acquired reference trouble type information relating to types of plumbing troubles and reference contract information relating to contract details of repair support for plumbing troubles is defined, the trained model having three or more levels of correlation with repair costs, the reference trouble type information and the reference contract information as input, and the repair costs as output, and giving priority to the reference trouble type information corresponding to the trouble type information acquired in the information acquisition step and the reference contract information corresponding to the contract information. A plumbing trouble repair cost estimation program that features:

2. In a plumbing trouble repair cost estimation program that estimates repair costs for plumbing troubles, an information acquisition step of acquiring trouble type information relating to the type of plumbing trouble and sound information recording sounds based on the plumbing trouble; and an estimation step of estimating repair costs by using a trained model in which a combination of previously acquired reference trouble type information relating to types of plumbing troubles and reference sound information relating to sounds caused by plumbing troubles and three or more levels of correlation with repair costs are defined, the trained model having the reference trouble type information and the reference sound information as inputs and the repair costs as outputs, and giving priority to the reference trouble type information corresponding to the trouble type information and the reference sound information corresponding to the sound information acquired in the information acquisition step. A plumbing trouble repair cost estimation program that features:

3. In a plumbing trouble repair cost estimation program that estimates repair costs for plumbing troubles, an information acquisition step of acquiring trouble type information relating to the type of plumbing trouble and location information relating to the location where the plumbing trouble occurred; and an estimation step of estimating repair costs by using a trained model in which a combination of previously acquired reference trouble type information relating to the type of plumbing trouble and reference location information relating to the location where the plumbing trouble occurred is specified with three or more levels of correlation with the repair costs, the trained model having the reference trouble type information and the reference location information as input and the repair costs as output, and giving priority to the reference trouble type information corresponding to the trouble type information and the reference location information corresponding to the location information acquired in the information acquisition step. A plumbing trouble repair cost estimation program that features:

4. In the information acquisition step, the trouble type information is acquired by capturing an image of the state of the plumbing trouble, In the estimation step, three or more levels of correlation between reference trouble type information obtained by capturing an image of the state of the plumbing trouble and the repair cost are defined.

2. The plumbing trouble repair cost estimation program according to claim 1,

5. The above correlations are composed of nodes in a neural network in artificial intelligence.

5. The plumbing trouble repair cost estimation program according to claim 1,

Images