Analysis system and analysis method

Abstract

To provide an analysis system configured to improve analysis accuracy even with little sample data.SOLUTION: A data creation system 2 calculates similarities between multiple pieces of aggregate data obtained by aggregating OD data 12 for each aggregation condition, and generates hierarchical cluster information which is a result of clustering processing which classifies the aggregate data into multiple clusters on the basis of the similarities. A processor 53 generates, on the basis of the hierarchical cluster information, for each condition of use, transportation user overall behavior data 20 obtained by supplementing supplement target data which is sample data in the condition of use, with sample data in a condition of use according to an aggregation condition of different aggregate data belonging to a cluster including aggregate data in an aggregate condition according to the condition of use. A transportation user behavior visualization system 3 analyzes behavior of users on the basis of the transportation user overall behavior data 20.SELECTED DRAWING: Figure 1

Description

Technical Field

[0006] , , , ,

[0001] The present disclosure relates to an analysis system and an analysis method.

Background Art

[0002] In recent years, analysis systems for analyzing the behavior of users who use transportation means such as railways have attracted attention. In this type of analysis system, for example, the behavior of users is analyzed using OD (Origin / Destination) data indicating the usage record of transportation means by each user and sample data indicating the behavior of each user.

[0003] The sample data can be generated, for example, from probe data indicating the position of a user at each time. However, unlike OD data that can be easily obtained by railway operators and the like, it is difficult to obtain a sufficient amount of probe data. For this reason, there is a problem that the analysis result of the analysis system is strongly affected by the bias of the sample data.

[0004] In contrast, Patent Document 1 discloses a technique for estimating a plurality of usage patterns in which a user uses a transportation means based on probe data, and complementing the sample data so that the congestion degree when assuming that all target users use the transportation means according to the usage pattern matches the actual congestion degree. According to this technique, since it is possible to increase the sample data, it is possible to improve the analysis accuracy.}]

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] [[ID=?]] It should be noted that there is a "?" in the above translation for the ID "24" and "42" in the original text where the specific content is not clear in the given text. If there are any inaccuracies due to this, please adjust according to the correct content.However, the technology described in Patent Document 1 has the problem of low versatility because it requires the actual level of congestion for each time and region in which usage patterns are estimated. Furthermore, if there are many different usage patterns, the amount of computation increases, and in some cases, the calculation may not converge.

[0007] The objective of the present invention is to provide an analytical system and analytical method that can improve analytical accuracy even when sample data is limited. [Means for solving the problem]

[0008] An analysis system according to one aspect of the present disclosure is an analysis system having a storage unit and a processing unit, wherein the storage unit stores population data showing the usage history of transportation by multiple users and behavioral sample data showing the behavior of sample individuals who are a part of the users, for each usage condition of the transportation; the processing unit calculates the similarity between multiple aggregated data obtained by aggregating the population data for each aggregation condition corresponding to the usage conditions, generates cluster information which is the result of a clustering process that classifies the aggregated data into multiple clusters based on the similarity; based on the cluster information, generates supplemented data for each usage condition by supplementing supplemental target data which is sample data of the usage condition with sample data of the usage condition according to the aggregation condition of another aggregated data belonging to the cluster which includes aggregated data of the aggregation condition corresponding to the usage condition; and analyzes the behavior of the users based on the supplemented data. [Effects of the Invention]

[0009] According to the present invention, it becomes possible to improve analytical accuracy even when sample data is limited. [Brief explanation of the drawing]

[0010] [Figure 1] This is a diagram showing an example of the configuration of an analysis system. [Figure 2]This is a diagram showing an example of the hardware configuration of a data creation system and a traffic user behavior visualization system. [Figure 3] This is a diagram showing an example of a route of a transportation facility. [Figure 4] This is a diagram showing an example of probe data. [Figure 5] This is a diagram showing an example of OD data. [Figure 6] This is a diagram showing an example of event information. [Figure 7] This is a diagram showing an example of operation information. [Figure 8] This is a flowchart for explaining an example of data creation processing. [Figure 9] This is a diagram showing an example of movement behavior sample data. [Figure 10] This is a diagram showing an example of time - band - aggregated OD data. [Figure 11] This is a diagram conceptually showing time - band - aggregated OD data. [Figure 12] This is a diagram showing an example of hierarchical cluster information. [Figure 13] This is a flowchart for explaining an example of supplementation processing. [Figure 14] This is a diagram for explaining an example of hierarchical supplementation processing. [Figure 15] This is a diagram for explaining another example of hierarchical supplementation processing. [Figure 16] This is a diagram for explaining an example of event supplementation processing. [Figure 17] This is a diagram for explaining an example of operation supplementation processing. [Figure 18] This is a diagram for explaining an example of time - band supplementation processing. [Figure 19] This is a diagram showing an example of sample supplementation record data. [Figure 20] This is a diagram showing an example of movement behavior sample data after supplementation. [Figure 21] This is a diagram showing an example of overall traffic user behavior data. [Figure 22] This is a diagram showing an example of an input screen. [Figure 23] It is a diagram showing an example of an output screen. [Figure 24] It is a diagram for explaining an example of the processing performed by the traffic user behavior visualization system according to Example 2. [Figure 25] It is a diagram for explaining an example of the processing performed by the traffic user behavior visualization system according to Example 3. [Figure 26] It is a diagram for explaining an example of the processing performed by the traffic user behavior visualization system.

Mode for Carrying Out the Invention

[0011] Hereinafter, examples of the present disclosure will be described with reference to the drawings.

Example

[0012] FIG. 1 is a diagram showing the configuration of an analysis system according to Example 1 of the present disclosure. The analysis system 1 shown in FIG. 1 is an apparatus that analyzes the behavior of traffic users who use a predetermined transportation facility. In the present embodiment, the predetermined transportation facility is a railway. However, the predetermined transportation facility is not limited to a railway, and may be, for example, a bus, a tram, an LRT (Light Rail Transit), or a combination thereof.

[0013] Returning to the description of FIG. 1, the analysis system 1 includes a data creation system 2 and a traffic user behavior visualization system 3.

[0014] The data creation system 2 is a creation apparatus that creates traffic user overall behavior data 20 that estimates the behavior of all users of a predetermined transportation facility based on input data 10 related to the predetermined transportation facility. Further, when creating the traffic user overall behavior data 20, the data creation system 2 performs a supplementation process of supplementing predetermined data (specifically, sample data described later), and creates sample supplementation record data 21 that is supplementation record data indicating the supplementation content by the supplementation process. The input data 10 includes, in the present embodiment, probe data 11, OD data 12, a planned train schedule 13, event information 14, and operation information 15.

[0015] Probe data 11 shows the location of samplers, who are a subset of transportation users, for each time (timestamp). OD data 12 is population data showing the actual usage of transportation by each transportation user. Planned timetable 13 shows the scheduled operation of a given transportation service. Event information 14 is data on events that affect the usage shown in OD data 12 (for example, events that take place around the route of a given transportation service). Operation information 15 shows the actual operation of a given transportation service.

[0016] The traffic user behavior visualization system 3 is a visualization device that analyzes the behavior of traffic users based on the overall traffic user behavior data 20 and sample supplementary record data 21 created by the data creation system 2, and creates visualization data that visualizes the analysis results. When creating visualization data, the traffic user behavior visualization system 3 may also use auxiliary data 22 as needed. Auxiliary data 22 may be, for example, point data (POI: Point of Interest).

[0017] Furthermore, the traffic user behavior visualization system 3 is connected to terminal 4 in a communicative manner and provides terminal 4 with input screens for inputting various information, output screens and display screens for presenting various information, etc. Terminal 4 is managed by, for example, a transportation operator that operates a designated transportation service, and is used by system users who utilize the traffic user behavior visualization system 3 (for example, a person in charge assigned to the planning department of the transportation operator).

[0018] Furthermore, the traffic user behavior visualization system 3 includes a display data extraction unit 31, a re-analysis implementation unit 32, a similarity condition calculation unit 33, a visualization data formatting unit 34, and a visualization unit 35. Detailed explanations of each part of the traffic user behavior visualization system 3 will be provided in Examples 2 to 4.

[0019] Figure 2 shows an example of the hardware configuration of the data creation system 2 and the traffic user behavior visualization system 3. As shown in Figure 2, the data creation system 2 and the traffic user behavior visualization system 3 each have a recording device 51, memory 52, processor 53, input device 54, and display device 55, and they are connected via a bus 56.

[0020] The recording device 51 is a storage unit that records data in a writable and readable format, and records a program that defines the operation of the processor 53, and various information used or generated by the processor 53 (input data 10, overall traffic user behavior data 20, sample supplementary recording data 21 and auxiliary data 22, and travel behavior sample data described later). The processor 53 is a processing unit that realizes functions according to the program by reading the program recorded in the recording device 51 into the memory 52 and executing it. For example, the processor 53 executes the processing performed by the data creation system 2. The processor 53 also realizes the display data extraction unit 31, the re-analysis implementation unit 32, the similarity condition calculation unit 33, the visualization data formatting unit 34, and the visualization unit 35 of the traffic user behavior visualization system 3. The input device 54 is a device that receives various information from the operator of the analysis system 1 and external devices (such as the terminal 4), and this information is used for processing by the processor 53. For example, input data 10 is recorded in the recording device 51 via the input device 54. The display device 55 is a device that displays various information.

[0021] Furthermore, the data creation system 2 and the traffic user behavior visualization system 3 may be implemented using, for example, the same physically identical information processing device (such as a server) having the configuration shown in Figure 2.

[0022] Figure 3 shows the route of a predetermined transportation system in this embodiment. In the example in Figure 3, the vehicle travels between stations ST0 and ST5. Station ST0 is in the suburbs, stations ST1 and ST2 are in residential areas, station ST3 is in a commercial area, and stations ST4 and ST5 are in an office district. Stations ST0 and ST4 are connected to routes of other companies that are different from the predetermined transportation system. A stadium where various events are held is located between stations ST3 and ST4.

[0023] Figure 4 shows an example of probe data 11. The probe data 11 shown in Figure 4 includes fields 111 to 114. Field 111 stores a user ID, which is identification information that identifies the sampler. Field 112 stores a timestamp indicating the time. Fields 113 and 114 store location information indicating the sampler's position at the time of the timestamp. Specifically, field 113 stores the latitude of the sampler's position, and field 114 stores the longitude of the sampler's position. Note that the probe data 11 may also have fields that store other information.

[0024] Figure 5 shows an example of OD data 12. The OD data 12 shown in Figure 5 includes fields 121 to 124. Field 121 stores the time. Fields 122 and 123 store route information indicating the route taken by the transport user using public transportation. Specifically, field 122 stores departure station information indicating the departure station from which the transport user boarded the vehicle, and field 123 stores arrival station information indicating the arrival station from which the transport user disembarked from the train. Field 124 stores the number of users, which is the number of transport users who used the route stored in fields 122 and 123 at the time stored in field 121. In this embodiment, the time stored in field 121 is the time when the transport user disembarked at the arrival station of that route.

[0025] Figure 6 shows an example of event information 14. The event information 14 shown in Figure 6 includes fields 141 to 143. Field 141 stores the date the event occurred. Field 142 stores relevant station information, indicating the station associated with the event. The relevant station is, for example, a station near the location where the event occurred. In the example in Figure 3, the location where the event occurred is the stadium, and the relevant stations are stations ST3 and ST4. Field 143 stores content information indicating the content of the event.

[0026] Figure 7 shows an example of operation information 15. The operation information 15 shown in Figure 7 includes fields 151 to 154. Field 151 stores the date on which an operation event related to the operation of a given transportation service occurred. An operation event is, for example, an event in which the operation deviates from the planned timetable. Field 152 stores the time period in which the operation event occurred. Field 153 stores operation details information indicating the content of the operation event. Field 154 stores the section (route) in which the operation event occurred.

[0027] Figure 8 is a flowchart illustrating an example of the data creation process performed in data creation system 2.

[0028] In the data creation process, first, the data creation system 2 analyzes the probe data 11 and performs a probe data basic analysis process to generate travel behavior sample data, which is behavior sample data that shows the content of the sampler's actions for each usage condition of public transportation (step S1). The usage conditions are defined by multiple items, and in this embodiment, they are defined by date, time of day, and route (departure station and arrival station). In this embodiment, the time of day is set in 30-minute intervals, but this is not the only example.

[0029] Next, the data creation system 2 performs an OD data aggregation process (step S2) to aggregate the OD data 12 according to aggregation conditions corresponding to the usage conditions and generate time-based aggregated OD data that includes multiple aggregated data. In this embodiment, the aggregation conditions are defined by the date and route included in the usage conditions. In other words, the data creation system 2 aggregates the OD data 12 for each date and route. Each aggregated data shows the time change (time-based value) of the number of users, which is the number of traffic users who used that route on that date, for each date and route. The aggregated data may also show the travel time, which is the average usage time using the route.

[0030] Then, the data creation system 2 performs a clustering process to classify the aggregated data into multiple clusters based on the similarity between the aggregated data included in the time-based aggregated OD data, and generates hierarchical cluster information by performing this process for each hierarchical level with a different total number of clusters (step S3). In the clustering process for each level, the number of clusters in that level may be determined, or the similarity threshold for dividing data into the same cluster may be specified for each level.

[0031] The similarity score is, for example, the weighted sum of the difference in the time taken for each aggregated data point and the similarity distance of the change in the number of users over time, i.e., α × (difference in time taken) + β × (similarity distance of change over time). Here, α and β are weight values. The similarity distance of change over time can be, for example, the Euclidean distance or DTW (Dynamic Time Warping).

[0032] Furthermore, the data creation system 2 generates supplemented travel behavior sample data, which is supplemented data, by performing a supplementation process to supplement the sample data within the travel behavior sample data according to the conditions for using transportation, based on the hierarchical cluster information (step S4). Supplementing sample data means supplementing the sample data to be supplemented, which is the sample data to be supplemented, with other sample data that has different usage conditions than the usage conditions of the sample data to be supplemented. The supplementation process will be described later with reference to Figures 13 to 18.

[0033] Finally, the data creation system 2 performs an expansion process to enlarge the sample data of behavioral behavior during movement after supplementation, and generates the expanded sample data of behavioral behavior during movement after supplementation as the overall behavioral data of all traffic users 20 (step S5), and then terminates the process. Specifically, the expansion process is a process that enlarges the sample data included in the sample data of behavioral behavior during movement after supplementation so that it matches the number of users in the OD data 12 that matches the usage conditions for each usage condition.

[0034] Figure 9 shows an example of sample data on movement behavior. The sample data 300 on movement behavior shown in Figure 9 includes fields 301 to 309.

[0035] Field 301 stores the user ID that identifies the sampler. Field 302 stores the date the sampler took action. Field 303 stores the day of the week the sampler took action. Field 304 stores the departure station from which the sampler boarded, and field 305 stores the arrival station from which the sampler disembarked. Field 306 stores the arrival time, which is the time of day the sampler arrived at the arrival station. Field 307 stores the main purpose of the sampler's action. Field 308 stores the arrival station sightseeing time, which is the time the sampler stayed around the arrival station. Field 309 stores information on the use of other modes of transport, indicating whether the sampler used modes of transport other than the specified modes of transport. In addition to fields 301 to 309, the travel action sample data 300 may also have fields that store other information, such as the places the sampler stayed (places visited). Note that each record of the travel action sample data 300 shown in Figure 9 is sample data.

[0036] The method for estimating each value of the travel behavior sample data 300 is not particularly limited. For example, the data creation system 2 can estimate the departure station, arrival station, time spent around the arrival station, and other transportation usage information by determining whether the travel route of the sampler shown in the probe data 11 resembles the shape of the route. The data creation system 2 can also determine the main purpose of the sampler's actions by clustering the purpose of their actions based on the time spent around the arrival station or the time of arrival, or by identifying the place of stay using POI data. If travel behavior sample data generated from the probe data 11 can be obtained from another system, the processing in step S1 can be skipped.

[0037] Figure 10 shows an example of time-based aggregated OD data. The time-based aggregated OD data 310 shown in Figure 10 includes fields 311 to 316.

[0038] Field 311 stores the date. Fields 312 and 313 store route information indicating the route of the transportation network. Specifically, field 312 stores the departure station, and field 313 stores the arrival station. Field 314 stores the arrival time, which is the time of day when the train arrived at the arrival station. Field 315 stores the day of the week. Field 316 stores the number of passengers, which is the number of passengers who boarded at the departure station in field 312 and arrived at the arrival station in field 313 during the arrival time in field 314 on the date in field 311. Here, time-based aggregated OD data 310 is generated for all combinations of date, time, and route.

[0039] Figure 11 is a conceptual diagram illustrating the time-based aggregated OD data 310 shown in Figure 10. In the example in Figure 11, aggregated data 31A showing the time-based change in the number of users is shown for each day for all routes, i.e., all combinations of arrival and departure stations.

[0040] Figure 12 shows an example of hierarchical cluster information. In the example in Figure 10, the hierarchy is divided into four levels, from 0 to 3, and the aggregated data belonging to a cluster in each level from 0 to 3 is indicated by a "○". As shown in Figure 10, the number of clusters decreases in the order of level 0, level 1, level 2, and level 3. Hereafter, a level with a small number of clusters will be referred to as a higher level. In other words, level 0 is the lowest level and level 3 is the highest level. Note that in the lowest level, all aggregated data is classified into a single cluster. The lowest level can also be considered as a state where clustering processing has not been performed.

[0041] Figure 13 is a flowchart that more specifically explains the supplementation process for supplementing the sample data of movement behavior in step S4 of Figure 8. The supplementation process is performed for each usage condition.

[0042] In the supplementation process, first, the data creation system 2 determines whether the number of target samples, which is the number of sample data to be supplemented, is equal to or greater than a threshold (step S101). In this embodiment, the threshold is set by the system user using the analysis system 1. The threshold may be set directly, or it may be determined in step S5 such that the magnification rate for expanding the target data to be supplemented is less than or equal to a predetermined tolerance value.

[0043] If the number of target samples is less than the threshold (Step S101: No), the data creation system 2 performs hierarchical supplementation processing to supplement the data based on the hierarchical cluster information (Step S102). Then, the data creation system 2 determines whether the number of target samples after supplementation is equal to or greater than the threshold (Step S103).

[0044] If the number of target samples after replenishment is below the threshold (step S103: No), the data creation system 2 performs event replenishment processing to replenish the target data based on the event information 14 (step S104). Then, the data creation system 2 determines whether the number of target samples after replenishment is above the threshold (step S105).

[0045] If the number of target samples after replenishment is below the threshold (Step S105: No), the data creation system 2 performs operation replenishment processing to replenish the target data based on the operation information 15 (Step S106). Then, the data creation system 2 determines whether the number of target samples after replenishment is above the threshold (Step S107).

[0046] If the number of target samples after supplementation is less than the threshold (Step S107: No), the data creation system 2 performs a time-based supplementation process to supplement the target data based on sample data from a time period different from the time period used for the supplementation target data (Step S108).

[0047] Furthermore, if the number of target samples is determined to be above the threshold in steps S101, S103, S105, and S107 (S101, S103, S105, and S107: Yes), and if step S108 is completed, the data creation system 2 generates the movement behavior sample data after replenishment and sample replenishment record data 21 indicating the content of the replenishment (step S109), and then terminates the process.

[0048] The replenishment process described above is merely an example and is not limited thereto. For example, the order in which the event replenishment process (step S104), operation replenishment process (step S106), and time zone replenishment process (step S108) are performed is not limited to the example above and can be changed as appropriate. Also, at least one of these processes may be omitted. Furthermore, the judgment processes such as steps S103, S105, and S107 may not be performed. In addition, even after the process in step S108 is completed, if the number of target samples is below the threshold, the data creation system 2 may change the replenishment conditions for replenishing the target data in each replenishment process and return to the process in step S102.

[0049] Figure 14 is a diagram illustrating an example of the hierarchical supplementation process in step S102 of Figure 13. In the example in Figure 14, the data to be supplemented is "sample data corresponding to users who rode from station ST1 to station ST4 during the 8:00 hour on February 4, 2022." The usage conditions at this time are expressed as "2022 / 2 / 4 8:00 ST1→ST4."

[0050] The data creation system 2 first searches the hierarchical cluster information for aggregated data with the aggregation condition "2022 / 2 / 4 ST1→ST4" corresponding to the usage condition "2022 / 2 / 4 8:00 ST1→ST4" (step S11). The data creation system 2 then determines the new current hierarchical level to be one level higher than the current hierarchical level (step S12). The initial value of the current hierarchical level is the lowest level, hierarchical level 0.

[0051] The data creation system 2 identifies other aggregated data included in the cluster to which the searched aggregated data belongs in the new current hierarchy (hierarchy 1 in the example in Figure 14) as target aggregated data (step S13). Then, the data creation system 2 supplements the data as supplemented data, which are sample data that meet the usage conditions according to the aggregation conditions of the target aggregated data (step S14).

[0052] Specifically, the data to be supplemented is sample data that includes the date and route specified in the aggregation conditions of the target aggregated data, and also includes the same time period as the usage conditions of the data to be supplemented. In the example in Figure 14, since the aggregation conditions of the target aggregated data are "2022 / 2 / 2 ST1→ST4" and "2022 / 2 / 3 ST1→ST4", the sample data for the usage conditions "2022 / 2 / 2 8:00 ST1→ST4" and "2022 / 2 / 3 8:00 ST1→ST4" becomes the data to be supplemented, and this data is supplemented as the data to be supplemented. In other words, in the example in Figure 14, sample data with different dates is supplemented as the data to be supplemented. This corresponds to, for example, supplementing data for the same time period on other weekdays that are thought to show a generally similar trend during commuting hours.

[0053] Figure 15 is a diagram illustrating another example of the hierarchical supplementation process in step S102 of Figure 13. In the example in Figure 15, the data to be supplemented is "sample data corresponding to users who rode from station ST1 to station ST3 in the 10 o'clock hour on February 5, 2022".

[0054] The data creation system 2 searches for aggregated data with the aggregation condition "2022 / 2 / 5 ST1→ST3" corresponding to the usage condition "2022 / 2 / 5 10:00 ST1→ST3" from the hierarchical cluster information, similar to the example in Figure 14 (step S21). The data creation system 2 determines the new current hierarchy to be one level higher than the current hierarchy (hierarchy 1 in the example in Figure 15) (step S22). In the new current hierarchy, the data creation system 2 identifies other aggregated data included in the cluster to which the searched aggregated data belongs as target aggregated data (step S23). Then, the data creation system 2 supplements the data to be supplemented, which is sample data of usage conditions corresponding to the aggregation condition of the target aggregated data, as supplemented data (step S24).

[0055] In the example in Figure 15, the aggregation conditions for the target aggregated data are "2022 / 2 / 4 ST2→ST3" and "2022 / 2 / 3 ST1→ST3". Therefore, sample data with the usage conditions "2022 / 2 / 4 10:00 ST2→ST3" and "2022 / 2 / 3 10:00 ST1→ST3" become the data to be supplemented, and this data is supplemented as the data to be supplemented. In other words, in the example in Figure 15, sample data with different dates and the same or slightly different departure stations are supplemented as the sample data to be supplemented. This corresponds to supplementing data from departure stations that are considered to have similar purposes, such as similar changes in the number of users over time and similar travel times.

[0056] The hierarchical supplementation process described in Figures 14 and 15 is executed by ascending the hierarchy until predetermined termination conditions are met. These termination conditions include, for example, that the number of aggregated data in the cluster to which the aggregated data belongs does not change when the hierarchy is raised, and that the current hierarchy is the highest level and there are no higher hierarchies. In the examples in Figures 14 and 15, all other aggregated data in the cluster to which the searched aggregated data belongs were identified as target aggregated data, but it is also possible to identify only the aggregated data that matches the predetermined supplementation conditions from among those other aggregated data as target aggregated data.

[0057] Figure 16 is a diagram illustrating an example of the event replenishment process in step S104 of Figure 13. In the example in Figure 16, the data to be replenished is "sample data corresponding to users who boarded a train from station ST1 to station ST3 in the 16:00 hour on March 4, 2022," and in the event replenishment process, the number of data points included in the cluster did not change even when the hierarchy was raised in the hierarchical replenishment process.

[0058] The data creation system 2 first checks the event information 14 to see if an event occurred on the date specified by the usage condition "2022 / 3 / 4 16:00 ST1→ST3" with the destination station being the relevant station (step S31). If the above event has occurred, the data creation system 2 adds sample data with the relevant station of that event as the destination station, and with the date, time, and departure station of the usage condition as the data to be added (step S32). In the example in Figure 16, sample data for the usage condition "2022 / 3 / 4 16:00 ST1→ST4" is added. If the above sample data does not exist, the data creation system 2 may add sample data from another day in which the same event occurred as the data to be added (step S33).

[0059] Figure 17 is a diagram illustrating an example of the operation supplementation process in step S106 of Figure 13. In the example in Figure 17, the data to be supplemented is "sample data corresponding to users who rode from station ST1 to station ST3 in the 9 o'clock hour on February 6, 2022," and the number of data included in the cluster did not change even when the hierarchy was raised in the hierarchical supplementation process.

[0060] The data creation system 2 first checks the operation information 15 to confirm whether an operation event occurred on the date of the usage condition to be supplemented, "2022 / 2 / 6 9:00 ST1→ST3" (step S41). If the above operation event has occurred, the data creation system 2 supplements the data with sample data for another time period in which the same operation event occurred on the date of the usage condition and at the departure and arrival stations (step S42). At this time, the data creation system 2 may supplement the data with sample data according to predetermined supplementation conditions. For example, the data creation system 2 may supplement a predetermined number of sample data starting with those having time periods different from the time period of the usage condition to be supplemented, or it may supplement with sample data that are included in a predetermined time range that includes the time period of the usage condition to be supplemented.

[0061] Figure 18 is a diagram illustrating an example of the time period supplementation process in step S108 of Figure 13. In the example in Figure 18, the hierarchy has been raised by one level through the hierarchy supplementation process, so the data to be supplemented are "sample data corresponding to users who boarded from station ST1 to station ST4 in the 8 o'clock hour on February 2, 2022", "sample data corresponding to users who boarded from station ST1 to station ST4 in the 8 o'clock hour on February 3, 2022", and "sample data corresponding to users who boarded from station ST1 to station ST4 in the 8 o'clock hour on February 4, 2022".

[0062] The data creation system 2 supplements the data to be supplemented with sample data that has the same date and route as the usage conditions of the data to be supplemented, but with different usage conditions for different time zones. In this case, the data creation system 2 supplements a predetermined number of sample data starting with those with time zones close to the time zone of the usage conditions of the data to be supplemented, or it supplements time zones included in a predetermined time range that includes the time zone of the usage conditions to be supplemented with sample data.

[0063] Figure 19 shows an example of sample supplementation record data 21. The sample supplementation record data 21 shown in Figure 19 includes fields 210 to 219.

[0064] Fields 210-213 store the usage conditions for the replenished sample data, which is the sample data that has been replenished. Specifically, field 210 stores the date of the replenished sample data, field 211 stores the departure station of the replenished sample data, field 212 stores the arrival station of the replenished sample data, and field 213 stores the time period of the replenished sample data. Field 214 stores the number of times replenishment was performed when replenishment was carried out on the replenished sample data. Field 215 stores the replenishment method, which is the type of replenishment process performed on the replenished sample data. Field 216 stores the post-replenishment conditions, which are the original usage conditions for the sample data included in the replenished sample data. Field 217 stores the replenishment condition trend, which shows the trend of the post-replenishment conditions. The replenishment condition trend shows, for example, the difference (different items and number) between the usage conditions of the replenished sample data and the usage conditions of the original sample data. Field 218 stores the current hierarchy. Field 219 stores the hierarchy remaining conditions, which indicate whether there is still sample data that can be replenished remaining in the current hierarchy.

[0065] Figure 20 shows an example of sample data on post-replenishment movement behavior. The post-replenishment movement behavior sample data 23 shown in Figure 20 includes fields 230-242.

[0066] Fields 230-233 store the usage conditions for the supplemented sample data. Specifically, field 230 stores the target date, which is the date of the supplemented sample data; field 231 stores the target departure station, which is the departure station of the supplemented sample data; field 232 stores the target arrival station, which is the arrival station of the supplemented sample data; and field 233 stores the target time period, which is the time period of the supplemented sample data. Field 234 stores the supplementation method. If the data is not supplemented, the supplementation method will be "original data". Field 235 stores the user ID of the sampler of the supplemented sample data. Fields 236-239 store the details of the actions taken by the sampler. Specifically, field 236 stores the date the sampler took the action; field 237 stores the departure station from which the sampler boarded; field 238 stores the arrival station from which the sampler disembarked; and field 239 stores the arrival time period, which is the time period when the sampler arrived at the arrival station. If the supplementation method is the original data, the same values ​​will be stored in fields 230-233 and fields 236-239. Field 240 stores the primary purpose, which is the objective of the sampler's actions. Field 241 stores the time spent around the arrival station, which is the sampler's stay around the arrival station. Field 242 stores information on other modes of transportation used, indicating whether the sampler used a mode of transportation other than the specified mode of transportation.

[0067] Figure 21 shows an example of overall traffic user behavior data 20. The overall traffic user behavior data 20 shown in Figure 21 includes fields 200 to 209.

[0068] Fields 200-203 store usage conditions. Specifically, field 200 stores the date, field 201 stores the departure station, field 202 stores the arrival station, and field 203 stores the time period. Field 204 stores the number of times the sample data for the usage conditions has been supplemented. Field 205 stores the supplementation method, which is the type of supplementation process performed on the sample data for the usage conditions. Field 206 stores the number of expanded sample data with the usage conditions as the number of users, which is the number of transportation users who used transportation under those usage conditions. Field 207 stores the main purpose of the transportation user. Field 208 stores the travel time around the arrival station of the transportation user. Field 309 stores information on other transportation used by the transportation user. The number of people in field 206 is expanded in step S5 of Figure 8 until it matches the number of people in OD data 12 for each usage condition.

[0069] Figure 22 shows an example of an input screen. The input screen 400 shown in Figure 22 has a data input section 401, a creation range input section 402, and a supplementary condition input section 403.

[0070] The data input section 401 is an area for inputting probe data 11, OD data 12, event information 14, and operation information 15 for creating overall traffic user behavior data 20. The planned timetable 13 is assumed to be predetermined.

[0071] The creation range input section 402 is an area for inputting the creation range for creating the overall traffic user behavior data 20. The creation range includes the creation period, which is a time-based range, and the creation region, which is a geographical range.

[0072] The supplemental condition input section 403 is an area for inputting threshold values, which are supplemental conditions. In the example shown in the figure, the supplemental condition input section 403 allows the user to choose whether to directly input a threshold value, input an acceptable value for the magnification of the sample data to set the threshold, or not to supplement the movement behavior sample data.

[0073] Figure 23 shows an example of an output screen. The output screen 410 shown in Figure 23 has a success / failure determination display unit 411, an output button 412, and a detailed display unit 413.

[0074] The success / failure determination display unit 411 is an area that indicates whether or not the creation of the overall traffic user behavior data 20 was successful.

[0075] The output button 412 is pressed when the created overall traffic user behavior data 20 is to be output. When the output button 412 is pressed, the data creation system 2 outputs the created overall traffic user behavior data 20 to the traffic user behavior visualization system 3.

[0076] The detailed display section 413 is an area that displays the supplementary content created by the supplementary processing used to create the overall traffic user behavior data 20. In the example shown in Figure 23, it includes the number of conditions 421, the average number of supplementary operations 422, the total number of supplementary methods 423, and the trend of differences in supplementary conditions 424.

[0077] The number of conditions (421) is the number of combinations of date, arrival station, departure station, and time zone included in the usage conditions of the supplemented sample data. The average number of supplementation operations (422) is the average number of times supplementation processing was performed on the sample data for each usage condition. The total number of supplementation methods (423) shows the number of times each type of supplementation processing was performed. The supplementation condition difference trend (424) shows the difference (different items) between the usage conditions of the sample data to be supplemented and the usage conditions of the sample data supplemented by the supplementation processing, and the number of such differences.

[0078] As described above, according to this embodiment, the processor 53 of the data creation system 2 calculates the similarity between multiple aggregated data sets obtained by aggregating OD data 12 for each aggregation condition, and generates hierarchical cluster information, which is the result of a clustering process that classifies the aggregated data into multiple clusters based on that similarity. Based on the hierarchical cluster information, the processor 53 generates overall traffic user behavior data 20 by supplementing the supplementary data, which is the sample data for each usage condition, with sample data for usage conditions corresponding to the aggregation conditions of other aggregated data belonging to a cluster that includes aggregated data for aggregation conditions corresponding to the usage condition. The processor of the traffic user behavior visualization system 3 analyzes the behavior of users based on the overall traffic user behavior data 20. Therefore, since it is possible to supplement sample data with data similar to that sample data, it is possible to improve the accuracy of the analysis even when there is little sample data.

[0079] Furthermore, in this embodiment, the processor 53 performs clustering processing for each hierarchical level with a different number of clusters, and replenishes the data to be replenished by changing hierarchical levels in order from the level with the most clusters until the number of data to be replenished reaches a threshold. As a result, the data is replenished in order from those with the highest similarity to the data to be replenished, making it possible to further improve the accuracy of the analysis.

[0080] Furthermore, in this embodiment, the processor 53 expands the number of data to be supplemented after supplementation until it matches the number of users corresponding to the aggregated data for each usage condition. This makes it possible to obtain sample data for all users of transportation services, thereby improving the accuracy of the analysis.

[0081] Furthermore, in this embodiment, the threshold for the data to be supplemented is set so that the magnification ratio of the data to be supplemented is below a predetermined tolerance value. This makes it possible to suppress the decrease in analysis accuracy caused by excessive magnification.

[0082] Furthermore, in this embodiment, even if predetermined termination conditions are met, if the number of data to be supplemented is below a threshold, the processor 53 supplements the data to be supplemented based on at least one of the event information 14 and the operation information 15. Therefore, even if suitable data to be supplemented cannot be determined from the OD data 12, it becomes possible to appropriately supplement the data to be supplemented.

[0083] Furthermore, in this embodiment, even if predetermined termination conditions are met, if the number of data to be supplemented is less than a threshold, the processor 53 supplements the data to be supplemented using sample data of usage conditions that include time periods different from those included in the usage conditions of the data to be supplemented. Therefore, even if it is not possible to determine suitable data to be supplemented from the OD data 12, it becomes possible to appropriately supplement the data to be supplemented.

[0084] In this embodiment, sample replenishment record data 21 is created that shows the replenishment details of the data to be replenished. This makes it possible for the user of the analysis system 1 to understand the replenishment details. [Examples]

[0085] Figure 24 is a diagram illustrating an example of the processing performed by the traffic user behavior visualization system 3 according to Example 2. In the example in Figure 24, the traffic user behavior visualization system 3 explains the process of analyzing and visualizing station usage as a measure of traffic user behavior.

[0086] The display data extraction unit 31 of the traffic user behavior visualization system 3 displays a display data specification screen 500 on the terminal 4 for inputting display conditions to specify the display data in the overall traffic user behavior data 20, and extracts display data from the overall traffic user behavior data 20 according to the display conditions entered using the display data specification screen 500. In the example in Figure 24, the display data specification screen 500 allows input of date, departure station, arrival station, and time period as display conditions.

[0087] The re-analysis unit 32 re-analyzes the display data extracted by the display data extraction unit 31 as needed and updates the display data. For example, the re-analysis unit 32 may use the display data and point data included in the auxiliary data 22 to analyze the places where the sample stayed, the number of people staying at each place, etc., and add this to the display data.

[0088] The similarity condition calculation unit 33 analyzes the usage conditions of the supplemented sample data, which are added to the sample data extracted as display data, as similarity conditions that are similar to the usage conditions of the display data, based on the overall traffic user behavior data 20 and the sample supplementation record data 21.

[0089] The visualization data formatting unit 34 generates display data 510, which is visualization data for visualizing the display data, based on the display data and similar conditions. The visualization unit 35 displays the display data 501 on the terminal 4. This can be used for considering construction sites for commercial facilities, locations for new stations, and considerations for stores to be attracted to facilities inside stations. [Examples]

[0090] Figure 25 is a diagram illustrating an example of the processing performed by the traffic user behavior visualization system 3 according to Example 3. In the example shown in Figure 25, the traffic user behavior visualization system 3 performs the following processing in addition to the processing according to Example 2.

[0091] The reanalysis unit 32 identifies abnormal data 511, which are sample data satisfying abnormal operation conditions related to events or operational phenomena, and normal data 512, which are sample data of usage conditions corresponding to the supplementation condition trends of the abnormal data, based on the overall traffic user behavior data 20 and the sample supplementation record data 21 from the displayed data. The reanalysis unit 32 analyzes the difference between the abnormal data 511 and the normal data 512. At this time, the reanalysis unit 32 may extract abnormal data from the displayed data where the supplementation method indicates an operation supplementation process.

[0092] Abnormal operation conditions indicate the date, time, and route in which the event or operational incident occurred. Usage conditions corresponding to supplementary condition trends are the usage conditions for sample data that have usage conditions shifted by the difference indicated by the supplementary condition trend from the abnormal time data.

[0093] The display screen 520 in Figure 25 is an example of the display screen in this embodiment, showing the total number of users of the destination, other transportation usage information, travel time around the arrival station, and differences in the destination between normal conditions ("no disruption") and abnormal conditions ("disruption present").

[0094] According to this embodiment, it becomes possible to visualize user behavior in the event of disruptions to transportation services, thereby supporting the development of measures to retain transportation users even when disruptions occur in public transport operations. [Examples]

[0095] Figure 26 is a diagram illustrating an example of the processing performed by the traffic user behavior visualization system 3 according to Example 4. In the example shown in Figure 26, the traffic user behavior visualization system 3 performs the following processing in addition to the processing according to Example 2.

[0096] The re-analysis unit 32 analyzes statistical values ​​of the number of users by primary purpose for each item that differs between the usage conditions of the supplemented sample data and the usage conditions of the original sample data, based on the supplementation condition trends of the overall traffic user behavior data 20 and the sample supplementation record data 21. The statistical values ​​are, for example, the mean, maximum, minimum, or deviation.

[0097] In supplementary condition trends, the date or day of the week often differs. The analysis results 530 shown in Figure 26 show the average number of users by route for each date. In the example in Figure 26, the primary purpose is "commuting," and the statistical value is "average."

[0098] The display screen 540 in Figure 26 is an example of the display screen in this embodiment, and shows a proposal based on the analysis results 530. In the example in Figure 26, even with the same number of users, shortening the time interval between transportation services is proposed for sections where the primary purpose is commuting, while widening the time interval between transportation services is proposed for sections where the primary purpose is shopping, etc.

[0099] In this embodiment, it becomes possible to make suggestions regarding the operation of transportation systems.

[0100] The embodiments of the Disclosure described above are illustrative for illustrative purposes and are not intended to limit the scope of the Disclosure to those embodiments only. Those skilled in the art can implement the Disclosure in various other forms without departing from the scope of the Disclosure. [Explanation of Symbols]

[0101] 1: Analysis system 2: Data creation system 3: Transportation user behavior visualization system 4: Terminal 31: Display data extraction unit 31A: Aggregated data 32: Re-analysis unit 33: Similarity condition calculation unit 34: Visualization data formatting unit 35: Visualization unit 51: Recording device 52: Memory 53: Processor 54: Input device 55: Display device

Claims

1. An analysis system having a memory unit and a processing unit, The storage unit stores population data showing the usage history of transportation services by multiple users, and sample data showing the behavior of a sample of the users, for each usage condition of the transportation services. The aforementioned processing unit, The similarity between multiple aggregated data sets obtained by aggregating the population data according to the aggregation conditions corresponding to the usage conditions is calculated. Cluster information is generated, which is the result of a clustering process that classifies the aggregated data into multiple clusters based on the similarity. Based on the cluster information, for each usage condition, the supplemented data is generated by adding the sample data of the usage condition, which is the sample data of the usage condition, to the supplemented data, which is the sample data of the usage condition, to the supplemented data, which is the sample data of the usage condition, which belongs to another aggregated data belonging to the cluster that includes aggregated data of the aggregation condition corresponding to the usage condition. Based on the supplemented data, the user's behavior is analyzed. The aforementioned usage conditions are defined by the date, time, and route on which the aforementioned transportation was used. The aforementioned aggregation conditions are defined by the date and route on which the aforementioned means of transport were used. The aggregated data shows, for each date and route, the change over time in the number of users (the number of users who used the route on that date) and the average usage time (the time taken to use the route). The aforementioned similarity is a weighted sum of the difference in required time in the aggregated data and the similarity distance of the change in the number of users over time. The processing unit generates supplementation record data indicating the supplementation details of the supplemented data, The supplementary record data shows the supplementary condition trend, which is the difference in usage conditions between the data to be supplemented and the data to be supplemented as the data to be supplemented. The processing unit is an analysis system that, based on at least one of operation information showing the operational performance of the transportation service and event information showing events that affect the usage performance of the transportation service and the supplementary record data, extracts abnormal operation data that satisfies abnormal operation conditions related to the operational performance or usage performance from the supplemented data, and normal operation data which is sample data of usage conditions corresponding to the supplementary condition trend of the abnormal operation data, and analyzes the difference between the abnormal operation data and the normal operation data.

2. The aforementioned processing unit, The clustering process described above is performed for each hierarchical level with a different total number of clusters. The analysis system according to claim 1, wherein the supplementation of the data to be supplemented is performed by changing the hierarchy sequentially, starting from the hierarchy with the most clusters, until the number of data to be supplemented reaches a threshold.

3. The analysis system according to claim 2, wherein the processing unit expands the data to be supplemented included in the supplemented data so that, for each usage condition, the number of users in the population data that matches the usage condition.

4. The analysis system according to claim 3, wherein the threshold is determined such that the magnification rate for expanding the supplemented data is less than or equal to a predetermined tolerance value.

5. The analysis system according to claim 2, wherein, even if a predetermined termination condition is met, if the number of data to be supplemented is less than the threshold, the processing unit supplements the data to be supplemented based on at least one of the operation information indicating the operation record of the transportation service and the event information indicating events that affect the usage record of the transportation service.

6. The aforementioned usage conditions are defined by the date, time, and route on which the aforementioned transportation was used. The analysis system according to claim 2, wherein if the number of data to be supplemented is less than the threshold even if a predetermined termination condition is met, the processing unit supplements the data to be supplemented by adding sample data with the same date and route as the usage conditions of the data to be supplemented, but with different time zones.

7. The analysis system according to claim 1, wherein the processing unit extracts sample data from the supplemented data according to predetermined display conditions as display data, and analyzes the usage conditions of the supplemented sample data that has been supplemented to the sample data extracted as display data as similar conditions that are similar to the usage conditions of the display data, based on the supplemented record data.

8. The aforementioned terms of use include multiple items, The supplementary record data shows the supplementary condition trend, which is the difference in usage conditions between the data to be supplemented and the data to be supplemented as the data to be supplemented. The analysis system according to claim 1, wherein the processing unit analyzes the number of users of the transportation service for each item that differs between the usage conditions of the data to be supplemented and the usage conditions of the data to be supplemented, based on the supplemented record data.

9. An analysis method using an analysis system having a memory unit and a processing unit, The storage unit stores population data showing the usage history of transportation services by multiple users, and sample data showing the behavior of a sample of the users, which is a subset of the users, for each usage condition of the transportation services. In the aforementioned processing unit, The similarity between multiple aggregated data sets obtained by aggregating the population data according to the aggregation conditions corresponding to the aforementioned usage conditions is calculated. Cluster information is generated, which is the result of a clustering process that classifies the aggregated data into multiple clusters based on the similarity. Based on the cluster information, for each usage condition, supplemented data is generated by adding sample data of the usage condition according to the aggregation conditions of another aggregated data belonging to the cluster, which includes aggregated data of the aggregation conditions according to the usage condition, to the supplemented data, which is the sample data of the usage condition. Based on the supplemented data, the user's behavior is analyzed. The aforementioned usage conditions are defined by the date, time, and route on which the aforementioned transportation was used. The aforementioned aggregation conditions are defined by the date and route on which the aforementioned means of transport were used. The aggregated data shows, for each date and route, the change over time in the number of users (the number of users who used the route on that date) and the average usage time (the time taken to use the route). The aforementioned similarity is a weighted sum of the difference in required time in the aggregated data and the similarity distance of the change in the number of users over time. The processing unit is instructed to generate supplementation record data indicating the supplementation details of the supplemented data. The supplementary record data shows the supplementary condition trend, which is the difference in usage conditions between the data to be supplemented and the data to be supplemented as the data to be supplemented. An analysis method comprising having the processing unit extract abnormal data that satisfies abnormal operation conditions related to the operation performance or usage performance from the supplemented data, and normal data which is sample data of usage conditions corresponding to the supplemented condition trend of the abnormal data, based on at least one of operation information showing the operation performance of the transportation system and event information showing events that affect the usage performance of the transportation system and the supplemented record data, and then analyzing the difference between the abnormal data and the normal data.

Images