Demand forecasting device, demand forecasting method, and demand forecasting program

Abstract

To enable the demand of new customers, which has been difficult to predict in the past, to be adequately predicted.SOLUTION: An AI prediction unit 110 predicts the number of visitors to a store in a target area based on information regarding geographic attributes in a GEO variable data file 104, search log information in a PSY variable data file 105, and people flow data in a DEM variable data file 106. In this case, the AI prediction unit 110 predicts the demand for new customers (the number of new customers visiting the store) based on people flow information in the target area, including information regarding land familiarization factors according to the frequency of visits to the target area.SELECTED DRAWING: Figure 6

Description

[Technical Field]

[0001] The present invention relates to an apparatus, method, and program for predicting the number of customers visiting a store such as a retail store or restaurant that is frequented by general consumers. [Background technology]

[0002] In the field of so-called marketing, various types of information are used to predict the number of customers (number of visitors) that will visit a retail store or restaurant. For example, Patent Document 1, which will be described later, discloses an invention relating to a prediction device, etc. In the invention disclosed in Patent Document 1, first, input information (search query) entered by a user and the user's location information at the time of entering the input information are acquired and stored in a storage unit. Next, based on the number of times the input information stored in the storage unit is entered, demand for the object corresponding to the input information in the area corresponding to the location information is predicted.

[0003] Furthermore, Patent Document 2, which will be described later, discloses an invention relating to an estimation device, etc. The invention disclosed in Patent Document 2 identifies attributes of a user at a specified location based on a search query used in a search and / or location data of a terminal, and estimates the user's potential demand at the specified location based on the identified attributes. In this way, the invention disclosed in Patent Document 1 predicts demand related to a search target, while the invention disclosed in Patent Document 2 estimates the user's potential demand at a specified location. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2019-53717 [Patent Document 2] Patent Publication No. 2021-99630 Summary of the Invention [Problem to be solved by the invention]

[0005] In the case of the inventions disclosed in Patent Documents 1 and 2, the factors are the user who performed the search and the location of the terminal at the time of the search. For example, whether the user is visiting the area for the first time or frequently visiting is not a factor to be considered. However, if the user has visited the area before, it is considered that the user is more likely to visit a store that they have visited before. Furthermore, if the user has visited the area before, for example, when searching for a restaurant and checking the search results, it is considered that the user is more likely to focus on restaurants near a location that the user knows. Therefore, in order to more accurately predict customer demand (such as the number of customers visiting a store), it is desirable to be able to accurately predict the demand of new customers who have never visited the area.

[0006] In view of the above, an object of the present invention is to make it possible to appropriately predict the demand of new customers, which has been difficult to do in the past. [Means for solving the problem]

[0007] In order to solve the above problem, the demand forecasting device of the invention described in claim 1 comprises: a first storage means for storing information on the geographic attributes of stores located in a destination area; a second storage means for storing search log information relating to stores in the destination area, the search log information being information relating to psychological attributes of stores in the destination area; a third storage means for storing information on the demographic attributes of the destination area, including information on the flow of people in the destination area, the information including local familiarity factor information according to the frequency of visits to the destination area; a fourth storage means for storing the actual number of customers collected by a terminal device installed in the store located in the destination area; The accumulated information on the geographical attributes, the search log information, and the people flow information of the target area including local familiarity factor information according to the frequency of visits to the target area; As explanatory variables, the accumulated number of actual store visitors is used as correct answer data to learn the Predict the number of visitors to stores in the target area The search log information is obtained by using search log information from an indirect search, and weighting the information on people who have less experience of visiting the target area as more important among the people flow information. Predictive measures and The present invention is characterized by comprising: [Effects of the Invention]

[0008] According to this invention, it is possible to perform prediction processing that emphasizes information on people who are unfamiliar with the area, which is part of local familiarity factor information in people flow information, which is information on demographic attributes. This makes it possible to appropriately predict demand from new customers, which was previously difficult. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a diagram illustrating an example of the configuration of a demand forecasting system according to an embodiment. [Figure 2] FIG. 2 is a diagram for explaining explanatory variables used in the demand forecasting system according to the embodiment. [Figure 3] FIG. 10 is a diagram illustrating an example of a geographic explanatory variable. [Figure 4] FIG. 10 is a diagram illustrating an example of a psychographic explanatory variable. [Figure 5] FIG. 10 is a diagram illustrating an example of a demographic explanatory variable. [Figure 6] FIG. 1 is a diagram illustrating a configuration example of a demand prediction device according to an embodiment. [Figure 7] 10 is a flowchart illustrating processing performed by the demand prediction device according to the embodiment. [Figure 8] FIG. 2 is a diagram for explaining phases (situations) when using the demand forecasting device according to the embodiment and the processing performed in each phase. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, with reference to the drawings, embodiments of the device, method, and program according to the present invention will be described. In the embodiments described below, for simplicity, an example will be described in which the number of visitors to a store located in a circular area (destination area) with a radius of 1000 m (meters) centered on a representative point of a specified station is predicted. Here, the circular area with a radius of 1000 m is assumed to be within walking distance from the station. Furthermore, the stores located in this area are retail stores, restaurants, etc., and do business when customers actually visit them.

[0011] [Example of demand forecasting system configuration] Fig. 1 is a diagram illustrating an example of the configuration of a demand forecasting system according to an embodiment. As shown in Fig. 1, the demand forecasting system according to this embodiment is configured by connecting a demand forecasting device 1, at least three explanatory variable providing server devices 3, 4, 5, and store systems 8(n), ... installed in each store, to a wide area network 2.

[0012] The demand forecasting device 1, which will be described in detail later, acquires and stores explanatory variables required to predict the number of store visitors from explanatory variable providing server devices 3, 4, and 5 connected to the wide area network 2, and predicts the number of store visitors using the stored explanatory variables. Note that the explanatory variables used in the demand forecasting device 1 do not mean constants, but rather input information input to the forecasting processing unit or forecasting program in order to predict the number of store visitors. The wide area network 2 is primarily the Internet, but also includes public switched telephone networks, optical telephone networks, mobile phone networks, LANs, wireless LANs, and the like that connect each device to the Internet.

[0013] The geographic explanatory variable providing server device 3 broadly provides information based on geographic attributes such as regional characteristics, climate, and regional population. In this embodiment, the geographic explanatory variable providing server device 3 provides information on store location factors and information on climatic factors of the area where the store is located. The psychographic explanatory variable providing server device 4 broadly provides information based on the psychological attributes of purchasers such as lifestyle, behavior, and values. In this embodiment, the psychographic explanatory variable providing server device 4 provides information on time factors and information on needs factors.

[0014] The demographic explanatory variable providing server device 5 provides information based on demographic attributes such as gender, age, and place of residence. In this embodiment, the demographic explanatory variable providing server device 5 provides information related to local familiarity factors and information related to factors related to the amount of stay in the vicinity. Details of the explanatory variables provided by each of these explanatory variable providing server devices 3, 4, and 5 will be described later.

[0015] A store system 8(n), ..., is composed of, for example, a POS (Point Of Sale) 6(n) and a store PC (Personal Computer) 7(n). The letter n in parentheses means an integer of 1 or greater, indicating that the system is configured for each store. In Figure 1, POS 6(n) is a so-called POS register, which records and accumulates information such as the product name, price, time of sale, and customer demographics (gender, age) at the time a product is sold (when payment is made), allowing for analysis, etc.

[0016] The store PC 7(n) has the function of receiving notification of predicted information on the number of store visitors from the demand forecasting device 1 and providing the information by displaying it on a display, and also providing information from the POS 6(n) to the demand forecasting device 1. Therefore, it is also possible to collect information on customer demographics, such as the actual number of store visitors, through the POS 6(n), and it becomes possible to provide the demand forecasting device 1 with the actual number of store visitors (correct answer data) relative to the predicted number of store visitors predicted by the demand forecasting device 1.

[0017] Customer demographics are generally information entered into the POS6(n) by the cashier, but in recent years it has become possible to automatically determine and enter this information using image recognition with a camera. Also, in stores that issue so-called membership cards, accurate information about customer demographics, such as gender and age, can be identified from the information on the membership card, and so-called repeat customer information, such as the interval between visits, can also be ascertained.

[0018] Furthermore, a store system 8(n) is not necessarily provided in every store. Some stores may only have a terminal device, such as a store PC 7(n), smartphone, or tablet PC, that communicates with the demand prediction device 1, and may not have a POS(n). In this case, even a conventional cash register can be used to input the number of customers who visit the store or to determine the actual number of customers from the number of payments (receipt issue numbers), etc., and such information can be provided to the demand prediction device 1 from the terminal device as correct answer data.

[0019] In the demand forecasting system of this embodiment, the demand forecasting device 1 is able to appropriately forecast demand from new customers, which has been difficult in the past, by devising information used as explanatory variables. Below, we will first explain the explanatory variables provided by each of the explanatory variable providing server devices 3, 4, and 5, and then explain the details of the demand forecasting device 1.

[0020] [Explanatory Variable Details] Fig. 2 is a diagram for explaining explanatory variables used in the demand forecasting system of the embodiment. As shown in the explanatory variable type column on the left side of Fig. 2, there are three types of explanatory variables used in the demand forecasting device 1: geographical (geographical attribute), psychographic (psychological attribute), and demographic (demographic attribute).

[0021] Geographic-based (geographic attribute-based) explanatory variables are information provided by the geographic-based explanatory variable providing server device 3 shown in Fig. 1, and the information that is the focus factor in this embodiment is information related to location factors and meteorological factors, as shown in Fig. 2. Specific examples of explanatory variables related to location factors include the number of competing stores around a station, the distance (travel distance) from the station to the store, and the time (travel time), as shown in the focus variable column in Fig. 2. Explanatory variables related to location factors can be acquired from a map information providing server device, as shown in the acquisition source column in Fig. 2.

[0022] Furthermore, a specific example of an explanatory variable related to meteorological factors is meteorological information about the area around a station, as shown in the variable of interest column in FIG. 2. The meteorological information includes not only the weather (sunny, cloudy, rainy, snowy, etc.) but also information such as temperature, humidity, and wind speed. The explanatory variables related to meteorological factors can be acquired from a meteorological information providing server device, as shown in the acquisition source column in FIG. 2. In this way, the geographic explanatory variable providing server device 3 may be made up of multiple server devices.

[0023] Figure 3 is a diagram illustrating an example of a geographic explanatory variable. Figure 3 shows a map of a circular area with a radius of 1000 m, centered around a representative point of a certain station. In Figure 3, the inner circle C1 has a radius of 500 m, and the outer circle C2 has a radius of 1000 m. In Figure 3, black circles indicate the location of, for example, a convenience store within circle C1, and white circles indicate the location of, for example, a convenience store in the area between circles C1 and C2.

[0024] As described above, the geographic explanatory variables used in the demand forecasting system of this embodiment are information that makes it possible to grasp the number of competing stores around a station, the distance (travel distance) from the station to each store, and the time (travel time). In other words, the map information providing server device, which is the geographic explanatory variable providing server device 3, does not simply provide map information. The map information providing server device has so-called facility POI (Point of Interest) information, including the locations of various stores and facilities as well as detailed store and facility information such as type and business hours, and is capable of grasping information of the type shown in FIG. 3.

[0025] Furthermore, the weather information for the area around the station is for the circular area shown in Figure 3, and in recent years, various weather companies have begun to provide weather information for smaller areas. For this reason, it is also possible to obtain weather information for the circular area shown in Figure 3 from the weather information providing server device. As a simple example, this makes it possible to grasp trends such as, for example, on rainy days, hot days, and days with high humidity, the number of customers at stores far from the station does not increase, but conversely, on sunny days with neither high temperature nor humidity, the number of customers at stores far from the station increases.

[0026] Psychographic (psychological attribute) explanatory variables are information provided by the psychographic explanatory variable providing server device 4 shown in FIG. 1, and the information that serves as the focus factor in this embodiment is information related to time period factors and needs factors, as shown in FIG. 2. Information related to time period factors and needs factors is information that can be acquired from search log information, as described above. A specific example of an explanatory variable related to time period factors is the time period during which searches were performed, as shown in FIG. 2, which is information indicating what kind of searches were performed and how many times for each time period. Furthermore, a specific example of an explanatory variable related to needs factors is the search words (search target) used in the search, as shown in FIG. 2.

[0027] Information on time-of-day factors and needs factors can be obtained from a search log server device of a company that provides a search engine, as shown in Figure 2. Note that information on time-of-day factors and needs factors can also include access information to websites operated by stores (known as owned media) or the companies that manage the stores.

[0028] There are two types of searches: natural searches, which are conducted regardless of the searcher's location, and local searches, which are conducted taking into account the location specified by the search words and the searcher's location. In other words, when a searcher enters a search word, a local search obtains search results focused on a specific area that varies depending on the places included in the search word and the searcher's location. Here, the searcher's location (current location) is information identified by a current location measurement function such as a Global Positioning System (GPS) installed in a communication device such as a smartphone that the searcher carries and uses to enter the search word and perform the search.

[0029] Furthermore, local searches include direct searches using store names or addresses, and indirect searches using generic names. For example, a search for "Lawson (registered trademark) near Tamachi Station" results in a direct local search, while a search for "convenience stores near Tamachi Station" results in an indirect local search. As shown in FIG. 2, the demand forecasting device 1 of this embodiment uses search log information for local searches, but particularly uses search log information for indirect searches. This is because in the case of an indirect search, the searcher often has not yet decided on the store they are going to visit, and there is a high possibility that they will become a new customer.

[0030] FIG. 4 is a diagram for explaining examples of psychographic explanatory variables. FIG. 4 shows an example of the aggregated log information of indirect searches within local searches. The log information can be aggregated at various time granularities, such as by time period, day, multiple days, week, or month. For simplicity's sake, the following example will be used to explain the aggregated log information of indirect searches within local searches performed in the morning (6:00-10:00), afternoon (10:00-14:00), and evening (18:00-21:00). Specifically, FIG. 4 shows information obtained by aggregating the log information of indirect searches within local searches performed using the name of a certain station and a generic name (category) of a store, such as "convenience store."

[0031] As can be seen from Figure 4, for example, in the morning, there are many searches for stores that offer quick shopping or a quick breakfast, such as convenience stores, coffee shops, and soba noodle stands. In the afternoon, there are many searches for stores for lunch, such as ramen shops, soba shops, and sushi restaurants, and in the evening, there are many searches for stores for drinking alcohol or having dinner, such as izakayas, Japanese restaurants, and other restaurants.

[0032] In this way, the psychographic explanatory variables used in the demand forecasting system of this embodiment make it possible to understand what types of stores searchers are looking for at various time granularities. Note that psychographic explanatory variables may also be formed using not only log information from indirect searches based on store categories, but also log information from indirect searches within local searches that use various search keywords along with station names, place names, addresses, etc.

[0033] Demographic (demographic attribute) explanatory variables are information provided by the server device 5 for providing demographic explanatory variables shown in Fig. 1, and the information that is the focus of attention in this embodiment is information related to local familiarity factors and nearby stay factors, as shown in Fig. 2. A specific example of an explanatory variable related to local familiarity factors is the amount of stay by frequency of inflow around a station, as shown in Fig. 2. Specifically, this information indicates how many times people in a specific area around a station have visited that area, and is information that can be ascertained from so-called people flow data (people flow information) managed by a mobile phone company, for example.

[0034] In addition to people flow data provided by mobile phone companies, various companies may collect, accumulate, and provide this data through application software provided by their own companies. Therefore, people flow data is not limited to data provided by mobile phone companies, and various types of available people flow data can be used. Of course, multiple people flow data provided by multiple providers can also be used.

[0035] Furthermore, as shown in Figure 2, a specific example of an explanatory variable related to nearby stay factors is the amount of stay by gender and age around a station, and this information can also be ascertained from the so-called people flow data managed by mobile phone companies. Therefore, the amount of stay by frequency of inflow around a station and the amount of stay by gender and age around a station can be obtained from a people flow data providing server device. Note that people flow data is information including identification information and current location information from mobile phone terminals such as smartphones, collected and managed by mobile phone companies. People flow data can be used to ascertain the behavior of people with certain attributes (gender, age, etc.) without identifying individuals.

[0036] By using this people flow data, it is possible to determine how many people are in a specific area around a station and how many people have visited the area before. In other words, it is possible to determine whether the person is visiting the area for the first time (no previous visitor), whether they have visited the area before but only a few times, or whether they visit the area frequently. This is possible because by accumulating information from mobile phone terminals over a long period of time, it is possible to determine the behavioral history of people in the area.

[0037] The demand forecasting device 1 of this embodiment uses information about local familiarity factors, i.e., the amount of stay around the station by frequency of inflow, as an explanatory variable. Furthermore, among the amount of stay around the station by frequency of inflow, information about people who have never visited before is treated as more important. This is because people who have never visited before are more likely to become new customers. Furthermore, after people who have never visited before, information about people who have only visited once may also be treated as important information. This is because in many cases, people have not yet decided which store to go to next, and so are more likely to become new customers.

[0038] FIG. 5 is a diagram illustrating an example of a demographic explanatory variable. FIG. 5 shows the results of tallying up people flow data on people who flowed into a predetermined circular area centered on a representative point of a certain station during each of the following time periods: morning (6:00-10:00), afternoon (10:00-14:00), and evening (18:00-21:00). Based on the people flow data, FIG. 5 identifies the number of people who flowed into the area according to their visit experience (number of visits). That is, in FIG. 5, each figure represents the number of people who flowed in, such as 100 people. Graph A shows the change in the inflow of people with no visit experience, graph B shows the change in the inflow of people with few visit experience, and graph C shows the change in the inflow of people with many visit experience.

[0039] As can be seen from Figure 5, for example, in the morning, many people with frequent visits flow in, and in the daytime there is not much difference depending on the difference in visitor experience, but in the evening, there is a large influx of people with no visitor experience. In this case, we can see that the evening time has a high potential for people with no visitor experience to become new customers.

[0040] As described above, the demander prediction system of this embodiment uses the geographic explanatory variables, psychographic explanatory variables, and demographic explanatory variables described with reference to Figures 2 to 5. Specifically, the demand prediction device 1 of this embodiment predicts the number of visitors to stores located in a specified area using the geographic explanatory variables, psychographic explanatory variables, and demographic explanatory variables described above. Note that the above-mentioned setting of the morning, afternoon, and evening time periods is merely an example, and the demand prediction device 1 described below can set various time periods for prediction.

[0041] [Configuration example of demand forecasting device 1] 6 is a diagram illustrating an example of the configuration of a demand forecasting device according to an embodiment. A connection terminal 101T constitutes a connection end to a wide area network 2, and a communication I / F (interface) 101 realizes a communication function via the wide area network 2. That is, the communication I / F 101 receives data sent to the device itself via the wide area network 2, converts the data into data in a format that can be processed by the device itself, and imports the data. The communication I / F 101 also converts data to be sent to a destination into data for transmission and transmits the data to the destination via the wide area network 2.

[0042] The control unit 102 is a microprocessor connected to a CPU (Central Processing Unit), ROM (Read Only Memory), RAM (Random Access Memory), and non-volatile memory, and controls each component of the demand forecasting device 1. The storage device 103 is an auxiliary storage device that includes a storage medium such as an SSD (Solid State Drive) and its driver, and is capable of writing, reading, changing, and deleting data from the storage medium. The storage device 103 stores and holds various data and programs required for processing, and is also used as a working area for temporarily storing intermediate results obtained in various processes.

[0043] The GEO variable data file 104 stores and holds the geographic explanatory variables acquired from the above-mentioned geographic explanatory variable providing server device 3. As described above, the geographic explanatory variables are information that can be used to determine the number of competing stores around the station, the distance from the station to the store, and the time, such as the information provided using FIG. 3, and meteorological information around the station. The information that can be used to determine the number of competing stores around the station, the distance from the station to the store, and the time, as well as meteorological information, are acquired retroactively from the present time and are stored and held. Note that if data is already stored in the GEO variable data file 104, only new data that has not been recorded will be acquired and additionally recorded.

[0044] The PSY variable data file 105 stores and holds the psychographic explanatory variables acquired from the psychographic explanatory variable providing server device 4. As described above, the psychographic explanatory variables are information such as the time period when a search was conducted and the search words used in the search, and provide the information described with reference to FIG. 4. Information such as the time period when a search was conducted and the search words used in the search is also acquired retroactively from the present time and stored and held. Note that if data is already stored in the PSY variable data file 105, only new data that has not been recorded will be acquired and additionally recorded.

[0045] The DEM variable data file 106 stores and holds demographic explanatory variables acquired from the demographic explanatory variable providing server device 5 described above. As described above, the demographic explanatory variables are information such as the amount of stay around a station by frequency of inflow and the amount of stay around a station by gender and age, and provide the information described using FIG. 5. Information such as the amount of stay around a station by frequency of inflow and the amount of stay around a station by gender and age is also acquired retroactively from the present time and stored and held. Note that if data is already stored in the DEM variable data file 106, only new data that has not been recorded will be acquired and additionally recorded.

[0046] The POS data file 107 receives and stores various POS data collected through the POS 6(n) installed in each store. The POS data from each store functions as so-called correct answer data used when adjusting the prediction process for the number of customers visiting the store. The GEO variable data file 104, PSY variable data file 105, and DEM variable data file 106 store and store the data provided by requests made to the explanatory variable providing server devices 3, 4, and 5 via the communication I / F 101 under the control of the control unit 102.

[0047] Furthermore, the GEO variable data file 104, PSY variable data file 105, DEM variable data file 106, and POS data file 107 are formed on predetermined SSDs. In this case, it is possible to form each file on one SSD, or to form each file on a different SSD. It is also possible to form the GEO variable data file 104, PSY variable data file 105, and DEM variable data file 106 on the same SSD, and the POS data file 107 on a different SSD.

[0048] The AI ​​(artificial intelligence) prediction unit 110 is a component implemented using so-called AI (artificial intelligence) technology, which uses the explanatory variables described above as input information to predict the number of visitors to stores located in a specified area. Here, AI (artificial intelligence) technology refers to a technology that artificially realizes human intelligence functions such as learning, inference, and judgment using a computer. Therefore, the AI ​​prediction unit 110 predicts the number of visitors to each store by learning and inferring using explanatory variables from the past few years or longer. The AI ​​prediction unit 110 also has the function of using the POS data (correct answer data) in the POS data file 107 to modify the algorithm and more accurately predict the number of visitors.

[0049] The AI ​​prediction unit 110 of the demand prediction device 1 of this embodiment includes four prediction units: a long-term trend prediction unit 111, a periodic fluctuation prediction unit 112, a short-term fluctuation prediction unit 113, and an irregular fluctuation prediction unit 114. The long-term trend prediction unit 111 predicts whether the number of visitors to each store will increase or decrease over a period of years, such as one year, three years, five years, or ten years. The periodic fluctuation prediction unit 112 sets a predetermined period, such as fluctuations throughout the year, within each month, or within each week, and predicts fluctuations in the number of visitors within that period. The short-term fluctuation prediction unit 113 predicts the number of visitors for each hour of the day, such as one hour, two hours, or three hours later. The irregular fluctuation prediction unit 114 predicts the possibility of large, irregular fluctuations in the number of visitors, for example.

[0050] The prediction results of the long-term trend prediction unit 111 are used in store consolidation and closure plans, such as store openings and closings. The prediction results of the periodic fluctuation prediction unit 112, short-term fluctuation prediction unit 113, and irregular fluctuation prediction unit 114 can be used in various cases. For example, they can be used in ordering raw materials, etc., to reduce waste. They can also be used to determine the timing of sales promotions and advertisements, to improve the effectiveness of sales promotions and advertisements. They can also be used to adjust store staffing, to prevent staff surpluses and shortages. They can also be used for dynamic pricing. Dynamic pricing is a system in which a certain standard price is set for a product or service and the price is adjusted as needed depending on the sales of that product or service. This makes it possible to appropriately determine the timing of price reductions for products and services.

[0051] The prediction result providing unit 120 realizes a function of providing the prediction results of the long-term trend prediction unit 111, the periodic fluctuation prediction unit 112, the short-term fluctuation prediction unit 113, and the irregular fluctuation prediction unit 114 to the store PC of the store with which the contract has a relationship via the communication IF 101. For example, information about the contract store (contractor), such as the email address of the contract store, is pre-registered in, for example, the storage device 103, and this information will be used.

[0052] [Processing of demand forecasting device 1] FIG. 7 is a flowchart illustrating the processing performed by the demand forecasting device according to the embodiment. The processing of the flowchart shown in FIG. 7 is executed by the control unit 102 of the demand forecasting device 1 at an appropriate timing when the number of visitors is predicted. The processing of the flowchart shown in FIG. 7 is executed. The control unit 102 first accesses the geographic explanatory variable providing server device 3 via the communication I / F 101, the connection terminal 101T, and the wide area network 2, and executes processing for collecting and storing geographic explanatory variables (step S1). The geographic explanatory variables collected in step S1 are stored in the GEO variable data file 104.

[0053] Next, the control unit 102 accesses the psychographic explanatory variable providing server device 4 via the communication I / F 101, the connection terminal 101T, and the wide area network 2, and executes collection and storage processing of the psychographic explanatory variables (step S2). The psychographic explanatory variables collected in step S2 are stored in the PSY variable data file 105. Next, the control unit 102 accesses the demographic explanatory variable providing server device 5 via the communication I / F 101, the connection terminal 101T, and the wide area network 2, and executes collection and storage processing of the demographic explanatory variables (step S3). The demographic explanatory variables collected in step S3 are stored in the DEM variable data file 106.

[0054] Thereafter, the control unit 102 controls the prediction units 111, 112, 113, and 114 of the AI ​​prediction unit 110 to execute a demand prediction process (a process for predicting the number of visitors to each store) (step S4). The demand prediction process of step S4 uses, in particular, information related to local familiarity factors and search log information related to indirect searches of local searches, thereby enabling appropriate prediction of demand from new customers (the number of new customers who will visit the store). After the demand prediction process of step S4, under the control of the control unit 102, the prediction result providing unit 120 functions to provide the prediction results to the contracted stores (step S5), and the process shown in FIG. 7 ends.

[0055] In this way, the demand forecasting device of this embodiment can perform a series of processes, such as collecting and storing explanatory variables, performing demand forecasting using the stored explanatory variables, and providing the forecast results to contracted stores (contractors). Each contracted store can use the provided forecast results to plan consolidation and closure, perform ordering, handle sales promotions and advertising, create work shifts, and handle dynamic pricing, as described above.

[0056] The processing performed as described above by the units of the demand forecasting device 1 functioning under the control of the control unit 102 of the demand forecasting device 1 of this embodiment is an application of an embodiment of a demand forecasting method according to the present invention. Also, the processing performed by the control unit 102 of the demand forecasting device 1 of this embodiment controlling the units is an application of an embodiment of a demand forecasting program according to the present invention.

[0057] Furthermore, in the process shown in FIG. 7, the process is described as being performed in the following order: (1) collection and accumulation of geographic variables → (2) collection and accumulation of psychographic variables → (3) collection and accumulation of demographic variables. However, this is not a limitation. The collection and accumulation process of these explanatory variables can be performed in various orders. Furthermore, the collection and accumulation process of each explanatory variable can also be performed in parallel. In other words, any order or parallel processing may be used as long as the explanatory variables of the target system can be collected and accumulated.

[0058] [Summary of the operation phases of the demand forecasting system] Fig. 8 is a diagram for explaining the phases (situations) when using the demand forecasting device 1 according to the embodiment and the processing performed in each phase. In Fig. 8, the input → conversion processing → output → correct answer data → utilization scene shown at the top indicates the phases when using the demand forecasting device 1, and the processing performed in each phase is shown below each phase.

[0059] In the input phase, the geographic explanatory variables, psychographic explanatory variables, and demographic explanatory variables are acquired by the demand forecasting device 1 and input into the AI ​​prediction unit 110 of the demand forecasting device 1. In the conversion processing phase, the long-term trend prediction unit 111, periodic fluctuation prediction unit 112, short-term fluctuation prediction unit 113, and irregular fluctuation prediction unit 114 of the AI ​​prediction unit 110 of the demand forecasting device 1 function to perform their respective intended prediction processes using the input explanatory variables.

[0060] In the output phase, the long-term trend prediction results, periodic fluctuation prediction results, short-term prediction results, and irregular fluctuation prediction results as predictions of the number of store visitors predicted by each section of the AI ​​prediction unit 110 of the demand prediction device 1 are output (output) and provided to the contracted store. In the correct data phase, information such as the actual number of store visitors, which is POS data collected at each store (contracted store), is fed back to the AI ​​prediction unit 110 of the demand prediction device 1. This allows the AI ​​prediction unit 110 to understand the cap (error) between the predicted number of store visitors and the actual number of store visitors, and corrects and changes the prediction algorithm etc. to adjust for this.

[0061] In the final usage scenario phase, the long-term trend forecast results are used by contracted stores to formulate store consolidation and closure plans. Additionally, the periodic fluctuation forecast results, short-term forecast results, and irregular fluctuation forecast results are used for raw material ordering, sales promotion and advertising, personnel adjustments, dynamic pricing, and other purposes.

[0062] [Effects of the embodiment] The demand forecasting device, demand forecasting method, and demand forecasting program of the above-described embodiments enable accurate forecasting of new customer demand, which has been difficult in the past, by focusing on general users who are unfamiliar with the area or who have not yet decided which store to visit in the area. In other words, explanatory variables that can capture factors such as lack of or limited previous visits or lack of familiarity with the surrounding area, as well as search log information from indirect searches within local searches, are used as explanatory variables. This enables more accurate planning of store consolidation and closure, ordering of raw materials, sales promotion and advertising, personnel adjustments, dynamic pricing, and other activities.

[0063] [Variations] In the above-described embodiment, the long-term trend prediction, periodic fluctuation prediction, short-term fluctuation prediction, and irregular fluctuation prediction are performed at the same timing, but this is not limited to this. For example, each prediction may be performed at a different timing, such as long-term trend prediction every six months, periodic fluctuation prediction every three months, short-term fluctuation prediction every day, and irregular fluctuation prediction every month. Also, short-term fluctuation prediction may be performed at shorter intervals, such as every hour or every three hours. In short, prediction processing can be performed at an appropriate timing depending on the type of prediction result.

[0064] Furthermore, in the above-described embodiment, the explanatory variables described with reference to Fig. 2 are used, but this is not limiting. Other explanatory variables may be used in addition to the explanatory variables described with reference to Fig. 2. Furthermore, explanatory variables other than information related to local familiarity factors and search log information of indirect searches in local searches may be replaced with different explanatory variables.

[0065] Furthermore, various explanatory variables may be used, and weighting may be applied to information related to local familiarity factors and search log information related to indirect searches in local searches when predicting demand. Furthermore, weighting may be applied to information related to people who are not familiar with the area among the information related to local familiarity factors. Various methods can be used to weight information related to local familiarity factors and search log information related to indirect searches in local searches. For example, measures can be taken such as increasing the coefficient by which information related to local familiarity factors and search log information related to indirect searches in local searches are multiplied, or increasing the number of times information related to local familiarity factors and search log information related to indirect searches in local searches are used. Of course, other methods may also be used.

[0066] [others] As can be seen from the above description of the embodiment, the function of the first accumulation means of the demand forecasting device of the claims is realized by the GEO variable data file 104 of the demand forecasting device 1 of the embodiment. Also, the functions of the second accumulation means and third accumulation means of the demand forecasting device of the claims are realized by the PSY variable data file 105 and the DEM variable data file 106 of the demand forecasting device 1 of the embodiment. Also, the function of the prediction means of the demand forecasting device of the claims is realized by the AI ​​prediction unit 110 of the demand forecasting device 1 of the embodiment.

[0067] As described above, the processing performed by the demand forecasting device 1 described using the flowchart in Fig. 7 is an application of an embodiment of the demand forecasting method according to the present invention. Furthermore, the program executed by the control unit 102 of the demand forecasting device 1 that performs the processing described using the flowchart in Fig. 7 is an application of an embodiment of the demand forecasting method according to the present invention. Note that the long-term trend prediction unit 111, cyclical fluctuation prediction unit 112, short-term fluctuation prediction unit 113, and irregular fluctuation prediction unit 114 that constitute the AI ​​prediction unit 110 of the demand forecasting device 1 can be realized as functions of the control unit 102 by the program executed by the control unit 102. [Explanation of symbols]

[0068] 1...demand forecasting device, 101T...connection terminal, 101...communication I / F, 102...control unit, 103...storage device, 104...GEO variable data file, 105...PSY variable data file, 106...DEM variable data file, 107...POS data file, 110...AI prediction unit, 111...long-term trend prediction unit, 112...periodic fluctuation prediction unit, 113...short-term fluctuation prediction unit, 114...irregular fluctuation prediction unit, 120...prediction result providing unit, 2...wide area network, 3...geographic explanatory variable providing server device, 4...psychographic explanatory variable providing server device, 5...demographic explanatory variable providing server device, 6(n)...POS, 7(n)...store PC, 8(n)...store system

Claims

1. a first storage means for storing information on the geographical attributes of stores located in a destination area; a second storage means for storing search log information relating to stores in the destination area, the search log information being information relating to psychological attributes of stores in the destination area; a third storage means for storing people flow information for the destination area, the information being information on demographic attributes for the destination area, including local familiarity factor information according to the frequency of visits to the destination area; a fourth storage means for storing the actual number of customers collected by the terminal devices installed in the stores located in the destination area; a prediction means for predicting the number of visitors to stores in the destination area by using, as explanatory variables, the accumulated information on the geographical attributes, the search log information, and people flow information in the destination area including local familiarity factor information according to the frequency of visits to the destination area, and learning using the accumulated actual number of store visitors as correct answer data, wherein the search log information is search log information from an indirect search, and weights, among the people flow information, information on people who have less experience of visiting the destination area as more important; A demand forecasting device comprising:

2. A demand forecasting device according to claim 1, The prediction means performs one or more of long-term trend prediction, periodic fluctuation prediction, short-term fluctuation prediction, and irregular fluctuation prediction. A demand forecasting device characterized by:

3. A demand prediction method used in a demand prediction device comprising: a first storage means for storing information on geographic attributes of stores located in a destination area; a second storage means for storing search log information on stores located in the destination area, which is information on psychological attributes of stores located in the destination area; a third storage means for storing people flow information in the destination area, which is information on demographic attributes of the destination area, including local familiarity factor information according to frequency of visits to the destination area; and a fourth storage means for storing actual numbers of store visitors collected by terminal devices installed in the stores located in the destination area, A demand forecasting method comprising a step of using the accumulated information on geographical attributes, the search log information, and people flow information for the destination area including local familiarity factor information according to the frequency of visits to the destination area as explanatory variables, and learning using the accumulated number of actual store visitors as correct answer data to predict the number of visitors to stores located in the destination area, wherein the search log information uses search log information from an indirect search, and weights, among the people flow information, information on people who have less experience of visiting the destination area as more important.

4. A demand forecasting program executed by a computer installed in a demand forecasting device comprising: a first storage means for storing information on geographic attributes of stores located in a destination area; a second storage means for storing search log information on stores located in the destination area, which is information on psychological attributes of stores located in the destination area; a third storage means for storing people flow information in the destination area, which is information on demographic attributes of the destination area, including local familiarity factor information according to the frequency of visits to the destination area; and a fourth storage means for storing the actual number of store visitors collected by terminal devices installed in the stores located in the destination area, A demand forecasting program characterized by executing a prediction step in which the program uses the accumulated information on geographical attributes, the search log information, and people flow information for the destination area including local familiarity factor information according to the frequency of visits to the destination area as explanatory variables, and learns using the accumulated number of actual store visitors as correct answer data to predict the number of visitors to stores located in the destination area, and in which, for the search log information, search log information from an indirect search is used, and, among the people flow information, information on people who have less experience of visiting the destination area is weighted as more important.

Images