JP2013152567A - Computing device, program and computing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a computing device capable of accurately computing the number of rail users on the basis of a ticket gate log, and a program and a computing method.SOLUTION: A computing device comprises: a storage section in which a passage log of a plurality of users passing through ticket gates, train time-table information, connection relation information of tracks, facility information in rail and statistical data on the number of users are stored; a generating section 304 for generating one or a plurality of behavior patterns associated with train boarding of the users associated with the passage log stored in the storage section or with stays of the users in station yards after passing through the ticket gates for a predetermined time period or longer on the basis of the passage log, the time-table information and the connection relation information of tracks stored in the storage section; a calculating section 305 for calculating probabilities that the users execute behaviors corresponding to one behavior pattern or more generated by the generating section 304 on the basis of the facility information, the passage log or the statistical data on the number of users stored in the storage section; and a computing section 306 for computing the number of users of the rail according to one behavior pattern or more generated by the generating section 304 on the basis of the probabilities calculated by the calculating section 305.

Description

  The present invention relates to a calculation device, a program, and a calculation method for calculating the number of railway users.

  In order to create a business plan for schedule revisions and installation of various facilities in the railway, the status of railway use is investigated. Therefore, there is a technique for grasping the use status of the railway or the number of users based on the record that the user passes the ticket gate (for example, Patent Document 1).

JP-A-11-165637

  However, the record that the user has passed through the ticket gates only reveals the entry time to the entry station, entry station, entry station premises, and entry time from the entry station premises. The detailed behavior of is unknown. For this reason, it is not possible to grasp the exact state of use of the railway simply by using the record that the user has passed through the ticket gate.

  The present invention has been made in view of such circumstances. The purpose is to provide a calculation device, a program, and a calculation method capable of accurately calculating the number of railway users based on the ticket gate record.

  One aspect of the disclosure according to the present application is a calculation device that calculates the number of railway users based on a passage record of ticket gates in a railway, a passage record including entrance times and entry times when a plurality of users pass the ticket gates, a train Operation time information, route connection information, railway equipment information and number of users statistical data storage, passage record, operation time information and route connection relationship stored in the storage unit A generating unit that generates one or a plurality of behavior patterns related to staying for a predetermined time or more in a ticket gate of a user's flight or a station ticket relating to a passage record stored in the storage unit based on the information of A calculation unit that calculates a probability that a user has executed an action corresponding to the action pattern generated by the generation unit based on the facility information stored in the storage unit, the passage record or the statistical data of the number of users; Based on the probability that the calculation unit has calculated, and a calculation unit for calculating a number of users train the generation unit corresponding to the generated behavior pattern.

  According to one aspect of the disclosure of the present application, it is possible to accurately calculate the number of railway users based on ticket gate records.

It is a block diagram which shows the structural example of a railway user number calculation system. It is a block diagram which shows the hardware structural example of a calculation apparatus. It is explanatory drawing which shows an example of the record layout of a ticket gate log table. It is explanatory drawing which shows an example of the record layout of the shortest movement time table. It is explanatory drawing which shows an example of the record layout of an operation record table. It is explanatory drawing which shows an example of the record layout of a route table. It is explanatory drawing which shows an example of the record layout of a station equipment scale factor table. It is explanatory drawing which shows an example of the record layout of a station table and a sales table. It is explanatory drawing which shows an example of the record layout of a railway utilization parameter | index table. It is a functional block diagram which shows the function structural example of a calculation apparatus. It is a flowchart which shows an example of the procedure of a railroad user number calculation process. It is explanatory drawing for demonstrating a definite log. It is explanatory drawing for demonstrating a definite log. It is explanatory drawing for demonstrating a definite log. It is explanatory drawing for demonstrating a definite log. It is explanatory drawing for demonstrating a route confirmation log. It is explanatory drawing for demonstrating a route confirmation log. It is explanatory drawing for demonstrating a route confirmation log. It is explanatory drawing for demonstrating a route confirmation log. It is explanatory drawing for demonstrating a route confirmation log. It is explanatory drawing for demonstrating a limitation log. It is explanatory drawing for demonstrating a limitation log. It is explanatory drawing for demonstrating a limitation log. It is explanatory drawing for demonstrating a limitation log. It is explanatory drawing which displayed as a list the probability calculation element with respect to each action pattern produced | generated based on the limitation log. It is explanatory drawing for demonstrating a non-limited log. It is explanatory drawing for demonstrating a non-limited log. It is explanatory drawing for demonstrating a non-limited log. It is explanatory drawing for demonstrating a non-limited log. It is a flowchart which shows an example of the procedure of case classification processing. It is a flowchart which shows an example of the procedure of a route decision log process. It is a flowchart which shows an example of the procedure of a non-limited log process. It is a flowchart which shows an example of the procedure of the total process based on a non-limited log. It is a flowchart which shows an example of the procedure of a limited log process. It is a flowchart which shows an example of the procedure of a similar log extraction process. It is a flowchart which shows an example of the procedure of a weight provision process.

  Hereinafter, embodiments will be described with reference to the drawings. The calculation device (computer) according to the present embodiment calculates the number of users of the transportation facility based on the ticket gate log. Transportation here includes railroads, subways, monorails, and the like. The ticket gate log is a pass record of ticket gates transmitted from the automatic ticket gate to the server via the network. The calculation device is a computer that can acquire a ticket gate log from a server, and is a PC (personal computer), a notebook PC, a workstation, a general-purpose computer, or the like. Hereinafter, a workstation that calculates the number of railway users will be described as an example of the calculation device. Of course, the calculation device may be a server that stores the ticket gate log transmitted from the automatic ticket gate.

FIG. 1 is a block diagram illustrating a configuration example of a railway user number calculation system.
The railway user number calculation system includes a plurality of automatic ticket gates 1, a server 2, and a calculation device 3. The automatic ticket checker 1, the server 2, and the calculation device 3 each have a communication function and are connected to each other via a network N. The network N is a dedicated communication line, a WAN (Wide Area Network), the Internet, a satellite communication line, a telephone communication line, or the like.

The automatic ticket gate 1 is a ticket gate installed at a station, and there are two types: a ticket gate for entering the ticket gate and a ticket gate for entering outside the ticket gate. The automatic ticket gate 1 may include a ticket gate for both entry and exit. Although one automatic ticket gate 1 is shown in FIG. 1, the automatic ticket gates 1 are installed at a plurality of stations, and the number thereof is plural.
The automatic ticket checker 1 reads information recorded on a boarding ticket, an admission ticket, a boarding pass ticket, a commuter pass, an electronic payment card or the like when the user passes the automatic ticket checker 1. The automatic ticket gate 1 transmits the information read from the ticket, the ticket gate passage date and time, the entrance / exit distinction, and its own identification information to the server 2 via the network N.
Information read from a boarding ticket etc. is recorded on a boarding ticket etc. by a magnetic recording surface, a semiconductor memory, etc., for example. The ticket gate passage date and time may be the date and time when the server 2 receives the ticket passage record from the automatic ticket gate 1.

  The server 2 is a workstation, for example, and has a database in its storage device. The server 2 receives various types of information from the automatic ticket gate 1 and stores the received information in the database as a ticket gate log. The ticket gate log here includes identification information such as a ticket, an entry station and an entry time, and an entry station and an entry time. The server 2 transmits the ticket gate log held in the database to the calculation device 3 via the network N.

When the automatic ticket checker 1 reads the information recorded on the commuter pass or the electronic payment card, the pass record includes the identification information of the commuter pass or the electronic payment card, respectively. The database of the server 2 includes a table in which attribute information of commuter passes and electronic payment cards is stored. Therefore, when the user passes the automatic ticket checker 1 using a commuter pass or an electronic payment card, the server 2 can extract attribute information such as the user's name, gender, date of birth, and the like.
The database of the server 2 includes a table in which information related to station facilities is stored. Therefore, when the user uses the equipment installed in the station premises by the electronic payment card, the server 2 can extract the attribute information of the used equipment in addition to the attribute information of the user.
When the server 2 extracts the user attribute information or the station facility attribute information, the server 2 also transmits the extracted attribute information to the calculation device 3 via the network N. Further, when the user uses the equipment installed in the station premises by the electronic payment card, the server 2 also transmits the use date and time of the equipment to the calculation device 3 via the network N.

  FIG. 2 is a block diagram illustrating a hardware configuration example of the calculation device 3. The calculation device 3 includes a CPU (Central Processing Unit, control unit, generation unit, calculation unit, calculation unit, reception unit, first determination unit, second determination unit, extraction unit, weight assignment unit, second generation unit, association Part) 31. The calculation device 3 includes a RAM (Random Access Memory) 32, a ROM (Read Only Memory) 33, a hard disk (storage unit) 34, a disk drive 35, a communication unit 36, a timer 37, a display unit 38, and an operation unit 39. . Each component of the calculation device 3 is connected via a bus 3b.

The CPU 31 controls each component of the calculation device 3. The CPU 31 reads the program 1P stored in the hard disk 34 and executes the program 1P.
The RAM 32 temporarily records work variables, data, and the like necessary in the process of the CPU 31. The RAM 32 is an example of a main storage device, and a flash memory, a memory card, or the like may be used instead of the RAM 32.

  The ROM 33 is, for example, a nonvolatile semiconductor memory or a read-only storage medium other than the semiconductor memory. The ROM 33 records a BIOS (Basic Input / Output System) executed by the CPU 31 when the calculation device 3 is activated, firmware, and the like.

  The hard disk 34 stores a program 1P executed by the CPU 31. The hard disk 34 may be mounted inside the calculation device 3 or may be placed outside the calculation device 3. The hard disk 34 is an example of an auxiliary storage device, such as a flash memory capable of recording a large amount of information, a CD (Compact Disc), a DVD (Digital Versatile Disk), a BD (Blu-ray Disc, registered trademark), or the like. It may be replaced by the optical disc 3a.

The hard disk 34 stores various master tables. The various master tables include a ticket gate log table 1T, a shortest travel time table 2T, an operation record table 3T, a route table 4T, a station facility scale factor table 5T, and a railway use index table 6T.
The hard disk 34 stores a work table (not shown).

The disk drive 35 reads information from the optical disk 3a, which is an external recording medium, and records information on the optical disk 3a.
The communication unit 36 is a modem or a LAN (Local Area Network) card, and is connected to the network N.

The timer 37 counts the date and time, and transmits the measured result to the CPU 31 as a signal.
The display part 38 has screens, such as a liquid crystal display and an organic EL (Electro-Luminescence) display, for example, and displays the various screens concerning the program 1P according to the instruction | indication from CPU31.

  The operation unit 39 includes input devices such as a keyboard, a mouse, and a touch panel on which a user performs various inputs. The operation unit 39 generates an input signal based on an operation by the user. The generated input signal is transmitted to the CPU 31 via the bus 3b.

Next, various master tables stored in the hard disk 34 will be described in detail.
FIG. 3 is an explanatory diagram showing an example of a record layout of the ticket gate log table 1T. The ticket gate log table 1T is a table for registering the ticket gate log that the automatic ticket gate 1 has read from the boarding ticket and transmitted to the server 2. The calculation device 3 receives the ticket gate log from the server 2 and registers the received ticket gate log in the ticket gate log table 1T.

  The ticket gate log table 1T includes columns of card ID, ticket gate passage data, and date / time. The card ID is an identification symbol such as a boarding ticket, an admission ticket, a boarding ticket, a commuter pass, or an electronic payment card read by the automatic ticket gate 1. The identification symbols of the ticket, admission ticket and boarding ticket are identification symbols assigned to each ticket by a ticketing machine or the like. The commuter pass identification symbol is an identification symbol given to the commuter pass by the railway operator's ticketing system. The identification symbol of the electronic payment card is an identification symbol given to the electronic payment card by the system of the business operator that issued the electronic payment card.

The ticket gate passage data is the station name of the station where the automatic ticket gate 1 is installed and the distinction of entry / exit. The date and time is the date and time when the user passes the automatic ticket gate 1. The date and time format is, for example, yyyy / mm / dd hh: mm: ss.
For example, it can be seen from the ticket gate log table 1T of FIG. 3 that the user who uses the railway with the card ID = ID001 took the following action. The user entered Station A at 12:30:14 on April 1, 20XX and entered from Station B at 15:39:52 on April 1, 20XX. The user entered the B station at 9:07:39 on April 2, 20XX, and entered the C station at 12: 58: 3 on April 2, 20XX.

  FIG. 4 is an explanatory diagram showing an example of the record layout of the shortest movement time table 2T. The shortest travel time table 2T is a table for registering the shortest travel time for each route when the section from the entrance station to the exit station is moved by train.

  The shortest travel time table 2T includes lines for routes, sections, and shortest travel times. The route is a route name. In addition, when the transfer of a route is possible in the middle station, the route before a transfer, the arrow, and the route after a transfer are stored in a route line like i line-> k line. Arrows indicate the transfer order. The section is a section where the user gets on the train. In the section, the entry station, the arrow, and the entry station are registered. Arrows indicate the direction of train movement. The shortest travel time is the shortest travel time for a route and a section. For example, when a user uses an express train, when the user arrives at a participating station in a shorter time than each blackout car, the required time when the user uses the express train is registered as the shortest travel time. In addition, if there is a halfway station where you can get off during the transfer from the entry station to the departure station, the time required for the user to move without registering for the shortest travel time is registered. .

  FIG. 5 is an explanatory diagram showing an example of a record layout of the operation record table 3T. The operation record table 3T is a table for registering the time when trains operated by date, route, and flight are operated. The hard disk 34 may store a timetable instead of the operation record table 3T. In such a case, the calculation device 3 refers to the train arrival and departure times using the timetable.

  The operation record table 3T includes columns of date, route, flight number, departure time from the first station, arrival / departure time on the way, and arrival time at the last station. The date is the date when the train operated. The date format is, for example, yyyy / mm / dd. The route is a route name. The flight name is the name of the train that operated. The first departure station departure time is the name of the first departure station and the departure time. The format format of the departure time is, for example, hh: mm. The station arrival / departure time is the name, arrival time, and departure time of the station where the service stops when there is a station that stops between the first station and the arrival station. The format format of arrival time and departure time is, for example, hh: mm. In the example of the operation record table 3T in FIG. 5, the arrival / departure time on the way is only one column. However, if there are a plurality of stations that are stopped on the way, the number of rows of arrival / departure times on the way is provided. The The terminal station arrival time is the name of the terminal station and the arrival time. The format format of the arrival time is, for example, hh: mm.

  If the arrival time and departure time at the halfway stop station are the same at the minute level, one of the arrival time and departure time is registered as the arrival / departure time on the way. The station arrival / departure time in the middle of the top record in FIG. 5 indicates a case where the arrival time and departure time are the same at the minute level. When the arrival time and the departure time are different at the minute level, both the arrival time and the departure time may be registered as the arrival / departure time on the way, or one of the times may be registered. If the flight is a train that does not have a stop station, the arrival / departure time train on the way is blank. The records on the second and third lines in FIG. 5 illustrate the case where the flight is a train that does not have a stop station on the way.

FIG. 6 is an explanatory diagram showing an example of a record layout of the route table 4T. The route table 4T is a table for registering connection relationships of routes. The route table 4T is a table that registers routes connected to each station, for example.
The route table 4T includes columns of station name, route 1, route 2, route 3,... Route n. The station name is the name of the station. Route 1, route 2, route 3,... Route n is the name of the route connected at the station of the station name. For example, at station A in FIG. 6, three lines, i line, j line, and k line, are connected. Therefore, at station A, it is possible to transfer trains between these three lines. For example, the line connected at the D station in FIG. 6 is only the j line. Therefore, there is no train transfer between routes at station D.

The train transfer here is a train transfer in the ticket gate. When the user passes through the automatic ticket gate 1 and enters the ticket gate, the calculation device 3 does not consider changing trains of other railway operators that are not involved in the railway user number calculation system.
Moreover, when the connection relation of a route can be extracted from the operation record table 3T or the timetable, the route table 4T may not be provided.

FIG. 7 is an explanatory diagram showing an example of a record layout of the station facility scale coefficient table 5T. The station equipment scale factor table 5T is a table for registering the scale or contents of equipment installed in the station. The station equipment scale factor table 5T is an example of a table for registering railway equipment. As other railway facilities, there are, for example, a wireless LAN spot in a train, a toilet, a public telephone, and a guidance display.
The station facility scale factor table 5T includes columns of station ID, station name, and number of stores in the ticket gate. The station ID is an identification symbol that identifies a station. The station name is the name of the station. The number of stores in the ticket gate is the number of stores installed in the ticket gate.

  FIG. 7 represents the scale of station facilities by the number of stores. However, the scale of the station equipment indicated by the station equipment scale factor table 5T may be the number of vending machines installed in the ticket gate, the total floor area of the waiting facility, and the like. Further, the scale of the station facilities may be the number of toilets installed, the number of advertisements attached, advertisement sales, the number of automatic external defibrillators (AEDs) installed, the number of wireless LAN spots installed, and the like. Furthermore, the scale of the station facility may be a combination of a plurality of facilities.

  FIG. 8 is an explanatory diagram showing an example of a record layout of the station table 51T and the sales table 52T. FIG. 8A is an explanatory diagram illustrating an example of a record layout of the station table 51T. FIG. 8B is an explanatory diagram illustrating an example of a record layout of the sales table 52T. The station table 51T in FIG. 8A and the sales table 52T in FIG. 8B are examples of tables that are combined to form the station facility scale factor table 5T. In the station table 51T and the sales table 52T in FIG. 8, the scale of the station facility is represented by the sales amount at the store in the ticket gate.

The station table 51T includes columns of station ID and station name. The station ID and the station name are the same as the station ID and the station name in the station facility scale factor table 5T of FIG. The sales table 52T includes columns of store ID, station ID, store name, and average sales. The store ID is an identification symbol that identifies the store in the ticket gate. The station ID is the same as the station ID in the station facility scale factor table 5T in FIG. The store name is the name of the store. The average sales is an average sales amount based on the total sales amount of the store in a predetermined period. The average sales are, for example, the total sales amount for the past year converted into an average amount per month. The unit of average sales is, for example, a yen.
The station ID of the station table 51T and the station ID of the sales table 52T are associated with each other. Therefore, by combining the station table 51T and the sales table 52T, a station equipment scale factor table 5T that represents the scale of the station equipment by the sales amount at the store in the ticket gate is constructed.

FIG. 9 is an explanatory diagram showing an example of a record layout of the railway use index table 6T. The railway use index table 6T is a table for registering the number of train passengers for each route measured in a macro manner.
The railway use index table 6T includes each column of route ID, route name, and daily average number of passengers. The route ID is an identification symbol that identifies the route. The route name is the name of the route and the traveling direction of the train. The traveling direction is distinguished by, for example, upstream and downstream. The daily average number of passengers is the average number of passengers per day based on the total number of passengers of all flights in a predetermined period. The daily average number of passengers is, for example, the total number of passengers in the past year converted to the average number of passengers per day for all flights on the route. The data that is the basis for the daily average number of passengers is the number of railway users estimated from the visual observation of the station staff at the limited observation positions, camera images, and the like.

In the railway usage index table 6T, the average number of passengers by flight, which is more detailed than the average number of passengers by route, may be registered.
Moreover, the railway use index table 6T may further include columns of station ID, station name, and daily average number of station users. The station ID and the station name are the same as the station ID and the station name in the station facility scale factor table 5T, respectively. The average number of station users per day is the average number of station users per day in a predetermined period. The daily average number of station users is, for example, the number of users for the past year passing through the automatic ticket gate 1 converted into the average number of users per day.
The railway use index table 6T may include data on the average number of passengers on each route or the average number of users on each station by season or by time of day.

  Next, the operation of the calculation device 3 will be described. The calculation device 3 obtains the probability that one user corresponding to one ticket gate log is basically in each flight or each station at the designated date and time. Then, the calculation device 3 uses the probability that the user corresponding to all the ticket gate logs to be processed was in each flight or each station at the designated date and time. The number of users who were on the bus and the number of users who were at each station are calculated.

  FIG. 10 is a functional block diagram illustrating a functional configuration example of the calculation device 3. Each function unit in FIG. 10 is executed by the CPU 31 operating based on the program 1P recorded on the hard disk 34. The calculation device 3 includes a log capturing unit 301, a reception unit 302, an extraction unit 303, a generation unit 304, a calculation unit 305, and a calculation unit 306.

The log capturing unit 301 acquires a ticket gate log from the server 2 via the communication unit 36, and stores the acquired ticket gate log in the ticket gate log table 1T. The accepting unit 302 accepts the date and time. The date and time is, for example, a second level. The calculation device 3 calculates the number of users in each flight and each station at the date and time received by the reception unit 302. Therefore, the date and time received by the receiving unit 302 is the date and time to be processed for calculating the number of railway users.
The CPU 31 displays an input screen (not shown) for the user to input the processing date and time on the display unit 38 based on the program 1P. The user inputs the target date and time for calculating the number of railway users on the input screen via the operation unit 39. The accepting unit 302 accepts the date and time input on the input screen.

  The extraction unit 303 extracts the ticket gate log corresponding to the user who has used the railway on the date and time received by the reception unit 302 from the ticket gate log table 1T. Specifically, the extraction unit 303 extracts from the ticket gate log table 1T the ticket gate log corresponding to the user who has entered the entrance station at the date and time received by the reception unit 302 and has not yet entered. Alternatively, the extraction unit 303 extracts from the ticket gate log table 1T the ticket gate log corresponding to the date and time received by the reception unit 302 between the entry time and the entry time.

  The generation unit 304 collates the ticket gate log extracted by the extraction unit 303 with the shortest travel time table 2T, the operation record table 3T, and the route table 4T, and the user who used the railroad at the date and time received by the reception unit 302. Generate behavior patterns. The behavior pattern here is a behavior type of how a user gets on which route on which route and how the user moves from the entrance station to the departure station. The user may have moved from the entry station to the departure station using only one flight. On the other hand, the user may get off at a station on the way, read a book at a bookstore on the way to the station, and then get on another flight and move from the entrance station to the departure station. Alternatively, the user may have boarded a flight on a certain route, changed to a flight on another route at a station, and moved from the entrance station to the departure station. The behavior patterns of users using such railways are limited to some extent because the number of routes and the number of trains are limited. Therefore, the generation unit 304 records the operation record of the train actually operated and the connection relation of the route, and the time when the user entered the station and when the user exited the train and exited the ticket gate. The log is collated, and a route, flight, and stopover pattern that is considered to be used by the user is generated.

Note that the generation unit 304 may or may not consider an action pattern that is considered to be a rare case. The rare case here is an action pattern when, for example, there are three stations that can be dropped off between the entrance station and the departure station, and the user has dropped off at all three stations. Alternatively, the rare case is an action pattern when the user gets off at two stations out of three stations on the way.
The rare case here refers to, for example, circular operation on the Yamanote Line in Tokyo, the Circle Line in London, the Subway Line 2 in Beijing, the L Orange Line in Chicago, and the like. Since the driving system of the circular operation operates in a circular shape, there are two possible behavior patterns for moving from the entry station to the departure station: outside and inside. In such a case, an action pattern that moves on a more detour route corresponds to a rare case.
Furthermore, the rare case here is a case where the entry station and the entry station coincide and the use of the train cannot be confirmed. In such a case, the generation unit 304 regards the user as having entered the station after staying at the entrance station, and generates an action pattern using only the station.
In addition, the stay in the station below is a stay in the ticket gate.

When there are only one user behavior pattern generated by the generation unit 304, which route of which route the user has boarded at the date and time received by the reception unit 302 or which station premises Is uniquely determined. However, when there are a plurality of user behavior patterns generated by the generation unit 304, which route of which route the user has boarded at the date and time received by the reception unit 302, or at which station of the station That is not unique.
Therefore, when there are a plurality of user behavior patterns generated by the generation unit 304, the calculation unit 305 calculates the probability that the user has performed the behavior corresponding to the behavior pattern for each behavior pattern. At that time, the calculation unit 305 uses a ticket gate log similar to the station equipment scale factor table 5T or the ticket gate log to be analyzed. A ticket gate log similar to the ticket gate log to be analyzed is extracted from the ticket gate log table 1T.
However, when the number of user behavior patterns generated by the generation unit 304 is greater than a predetermined number, the calculation unit 305 uses the railway usage index table 6T to determine the probability that the user has boarded the train on each route and the station premises. The probability that there was a user in is calculated.

  Based on the probability calculated by the calculation unit 305 for the user's behavior pattern, the calculation unit 306 adds up the number of users who have used the railway on the date and time received by the reception unit 302. In other words, the calculation unit 306 totals how many users are in which flights and in which stations on the date and time received by the reception unit 302.

  In the above description, the extraction unit 303 and the calculation unit 305 extract the ticket gate log from the ticket gate log table 1T. However, the extraction unit 303 and the calculation unit 305 may sequentially acquire the ticket gate log from the database of the server 2 via the network N. In such a case, the ticket gate log table 1T and the log capturing unit 301 may be omitted.

FIG. 11 is a flowchart illustrating an example of a procedure of a railway user number calculation process. In FIG. 11, it is assumed that the process of acquiring the ticket gate log from the server 2 and registering the acquired ticket gate log in the ticket gate log table 1T has been completed.
The CPU 31 receives the processing target date and time (step S101). The CPU 31 extracts from the ticket gate log table 1T the ticket gate log corresponding to the received processing target date and time between the entry time and the departure time (step S102). That is, the CPU 31 extracts from the ticket gate log table 1T the ticket gate log corresponding to the user who has been using the railway at the received processing target date and time. The CPU 31 selects one unprocessed ticket gate log from the extracted ticket gate logs (step S103).

  CPU31 performs a case classification process (step S104). The case classification process is a process of dividing the ticket gate log according to the user's action pattern and calculating the probability corresponding to the action pattern corresponding to the case-by-case ticket log. The CPU 31 determines whether or not there is an unprocessed ticket gate log (step S105). When it is determined that there is an unprocessed ticket gate log (step S105: YES), the CPU 31 returns the process to step S103. When it is determined that there is no unprocessed ticket gate log (step S105: NO), the CPU 31 totals the number of railway users at the accepted date and time based on all behavior patterns for which probabilities have been calculated (step S106). The process ends.

By the case classification process in step S104, the CPU 31 classifies the ticket gate log into four types. These four types of ticket gate logs are referred to as a confirmation log, a route confirmation log, a limited log, and a non-limited log, respectively.
The confirmation log is a ticket gate log that can uniquely determine a user's action pattern. The confirmation log is, for example, a ticket gate log having information for uniquely determining the route and flight of the train on which the user has boarded.
The route confirmation log is a ticket gate log having information for uniquely determining the route of the train on which the user has boarded. The flight of the train on which the user boarded is not uniquely determined from the route confirmation log.
The limited log is a ticket gate log having information that can limit the number of user behavior patterns to a predetermined number or less, although the route and flight of the train on which the user has boarded are not uniquely determined.
The unrestricted log is a ticket gate log having information in which both the route of the train on which the user has boarded and the flight are not uniquely determined and the number of user behavior patterns cannot be limited to a predetermined number or less.

  Hereinafter, specific logs will be described with respect to the fixed log, route fixed log, limited log, and non-limited log. A flow of a process for generating a behavior pattern and a process for calculating a probability after the case introduction of the confirmed log, the route confirmed log, the limited log, and the non-limited log will be described with reference to a flowchart.

  12, FIG. 13, FIG. 14 and FIG. 15 are explanatory diagrams for explaining the confirmation log. FIG. 12 is an explanatory diagram showing a record example of the ticket gate log table 1T corresponding to the confirmation log. FIG. 13 is an explanatory diagram showing a record example of the shortest movement time table 2T. FIG. 14 is an explanatory diagram showing a record example of the operation record table 3T. In FIG. 14, the train arrival / departure time train is omitted. FIG. 15 is an explanatory diagram showing an example of a route map. FIG. 15 is a route map visualizing the information in the route table 4T.

From FIG. 12, it can be seen that the user has entered A station and has exited from B station. The elapsed time in the meantime is 10 minutes 48 seconds. FIG. 15 shows that there are two routes connecting the A station and the B station, i line and j line. However, from the comparison between the elapsed time and the shortest movement time in FIG. 13, the minimum movement time 12 minutes for the j-line is longer than the elapsed time, so that it is determined that the user has used the i-line. In addition, from the comparison of the date and time in FIG. 12 with the departure time and arrival time at the first station in FIG. 14, the user took Y flight on i line and arrived at B station without getting off at C station. I understand. This is because the date and time of the ticket gate log in FIG. 12 does not match the operation records of the X and Z flights on the i line. Therefore, the user's behavior pattern corresponding to the ticket gate log in FIG.
Station A entrance / stay (12: 31: 14-12: 32) → Y boarding (12: 32-12: 41) → Station B stay / exit (12: 41-12: 42: 02)

  The ticket gate log in FIG. 12 is a confirmation log because a user's behavior pattern can be uniquely determined. When the ticket gate log is a confirmation log, the calculation device 3 sets 1 as the probability of the calculation result in the action pattern corresponding to the confirmation log.

  16, FIG. 17, FIG. 18, FIG. 19 and FIG. 20 are explanatory diagrams for explaining the route determination log. FIG. 16 is an explanatory diagram showing a record example of the ticket gate log table 1T corresponding to the route confirmation log. FIG. 17 is an explanatory diagram of a record example of the shortest movement time table 2T. FIG. 18 is an explanatory diagram of a record example of the operation record table 3T. FIG. 19 is an explanatory diagram of a record example of the station facility scale coefficient table 5T. FIG. 20 is an explanatory diagram showing an example of a route map. FIG. 20 is a route map visualizing the information in the route table 4T.

  It can be seen from FIG. 16 that the user has entered A station and has exited from B station. From the route map of FIG. 20, the route connecting the A station and the B station is only the i line. Therefore, it can be seen that the user's behavior pattern corresponding to the ticket gate log in FIG. 16 is a pattern of moving from the A station to the B station using the i line. The ticket gate log in FIG. 16 is a route confirmation log because the use route in the user's behavior pattern is uniquely determined.

  From FIG. 16, the elapsed time from when the user enters the A station to the exit from the B station is 32 minutes 4 seconds. On the other hand, the shortest movement time in FIG. 17 is 9 minutes. The difference between the elapsed time of 32 minutes and 4 seconds and the shortest travel time of 9 minutes is 23 minutes and 4 seconds, and the user may have spent more than 20 minutes at the store in the station. There is a possibility that

  Accordingly, when the difference between the elapsed time and the shortest travel time is longer than the predetermined time, the calculation device 3 generates an action pattern including staying at the station using the station equipment. Here, it is assumed that the calculation device 3 compares the difference between the elapsed time and the shortest movement time with a threshold value of 20 minutes. The calculation device 3 compares 23 minutes and 4 seconds with the threshold value of 20 minutes, and determines that the difference between the elapsed time and the shortest movement time is longer than the threshold value. And the calculation apparatus 3 produces | generates the action pattern which the user stayed for more than predetermined time in any one of A, C, and B station from the route map of FIG. For example, when the predetermined time for staying at the station is set to 20 minutes or more, the calculation device 3 calculates the stay for 20 minutes or more at the A station, the stay for 20 minutes or more at the C station from the route map of FIG. Attempts to generate three behavior patterns for stays over 20 minutes at

  Next, the calculation device 3 lists a train boarding pattern when the station stay exceeds 20 minutes. When the ticket gate log in FIG. 16 and the operation record in FIG. 18 are collated, the stay at station A exceeds 20 minutes only when the user uses the XZ flight. The stay at C station exceeds 20 minutes only when the user gets on Y flight, gets off at C station, and gets on XZ flight from C station. The stay at station B exceeds 20 minutes only when the user uses flight Y. The calculation device 3 calculates the probability corresponding to the number of stores in the ticket gate of FIG. 19 for the three estimated behavior patterns. Specifically, the calculation device 3 calculates the probability for the behavior pattern when staying at each station based on 3: 6: 4, which is the ratio of the number of stores in the ticket gates of A station, B station, and C station. To do.

The above three behavior patterns and probabilities are shown below.
(1) Station A entrance / stay (12: 31: 14-12: 53) → XZ boarding (12: 53-13: 02) → Station B stay / exit (13: 02-13: 03: 18)
The probability of staying at station A is (100 × 3) / (3 + 6 + 4) = 23.08%.
(2) Station A entrance and stay (12: 31: 14-12: 32) → Y flight (12: 32-12: 36) → C station stay (12: 36-12: 57) → XZ boarding ( 12: 57-13: 02) → Station B stay and participation (13: 02-13: 03: 18)
The probability of staying at station C is (100 × 6) / (3 + 6 + 4) = 46.15%.
(3) Station A entrance / stay (12: 31: 14-12: 32) → Y boarding (12: 32-12: 41) → B station stay / exit (12: 41-13: 03: 18)
The probability of staying at station B is (100 × 4) / (3 + 6 + 4) = 30.77%.

  In the case described above, a plurality of behavior patterns in which the stay time at the station exceeds 20 minutes are generated for each of the A station, the C station, and the B station. On the other hand, when there is no station where the stay time exceeds 20 minutes, the calculation device 3 regards it as an action pattern without staying for a predetermined time or more at the station, and uses a ticket gate log similar to the ticket gate log to be processed, The probability corresponding to each behavior pattern is calculated. The process using the similar ticket gate log will be described later in the explanation of the limited log case.

In the case of the route determination log described above, the number of stores in the ticket gate is used as a criterion for probability calculation, but the number of stores in the ticket gate and the sales amount of the store may be used as a criterion for probability calculation. Alternatively, only the sales amount of the store may be used as a criterion for probability calculation. The station equipment scale factor used as a reference for calculating the probability may be a numerical value corresponding to the number or contents of the equipment that can be used in the ticket gate, and may be a numerical value related to other equipment other than the store in the ticket gate.
In addition, when an electronic payment card is used to pass through the automatic ticket gate 1 and is used at a store in the ticket gate of the station where the electronic payment card is located, the electronic payment card is used at the station where the user stays. You may decide in the station where the store in a ticket gate is located.

  In the case of the route determination log described above, a plurality of behavior patterns are generated. However, there are cases where only one behavior pattern is generated. In such a case, the ticket gate log is classified as a fixed log, and the probability set for one action pattern is 1.

In the case of the route determination log described above, the difference between the elapsed time and the shortest travel time was compared with a threshold of 20 minutes. The threshold value may be described as a constant in the program 1 or may be received from the user by the calculation device 3. When the calculation device 3 accepts a threshold value to be compared with the difference between the elapsed time and the shortest travel time from the user, the CPU 31 displays an input screen (not shown) for that purpose on the display unit 38, and the user passes through the operation unit 39. A threshold value is input from the input screen.
Further, when generating an action pattern in which a user stays at a station for a predetermined time or longer, the predetermined time may be the same time as the threshold value or a different time. When the predetermined time is set to a time different from the threshold, the predetermined time may be described in the program 1 or the calculation device 3 may receive it from the user.

  In the case of the route determination log described above, the train boarding pattern when the station stay exceeds 20 minutes is one for each of the cases when staying at the A, C, and B stations. However, when staying at station C for 20 minutes or more, for example, when there are a plurality of flights that can be boarded, the calculation device 3 may assign a probability to a plurality of train boarding patterns according to the scale or content of the facilities of the flights.

  21, FIG. 22, FIG. 23, and FIG. 24 are explanatory diagrams for explaining the limited log. FIG. 21 is an explanatory diagram showing a record example of the ticket gate log table 1T corresponding to the limited log. FIG. 22 is an explanatory diagram showing a record example of the shortest movement time table 2T. FIG. 23 is an explanatory diagram of a record example of the operation record table 3T. FIG. 24 is an explanatory diagram showing an example of a route map. FIG. 24 is a route map visualizing the information in the route table 4T.

  From FIG. 21, it can be seen that the user has entered A station and has entered B station. From the route map of FIG. 24, the route connecting the A station and the B station is the i-line, the j-line, and the i-line and the k-line by transfer at the C-station. The elapsed time based on the ticket gate log in FIG. 21 is 21 minutes 47 seconds, which is not significantly different from the shortest travel time of each route according to FIG. Since there is not one action pattern that can be generated, the ticket gate log in FIG. 21 is not a fixed log. Further, since there is not one route corresponding to the behavior pattern, the ticket gate log in FIG. 21 is not a route confirmation log. The calculation device 3 collates the ticket gate log in FIG. 21 and the operation record in FIG. 23, and generates the following five patterns as action patterns that match the operation record.

(1) Station A entrance / stay (12: 31: 14-12: 32) → boarding X flight (12: 32-12: 41) → station B stay / exit (12: 41-12: 53: 01)
This behavior pattern uses i-line.
(2) Station A entrance / stay (12: 31: 14-12: 40) → boarding Y flight (12: 40-12: 49) → station B stay / exit (12: 49-12: 53: 01)
This behavior pattern uses i-line.
(3) Station A entrance / stay (12: 31: 14-12: 32) → boarding X (12: 32-12: 37) → staying at station C (12: 37-12: 38) → boarding BB ( 12: 38-12: 46) → Stay at B station and participate (12: 46-12: 53: 01)
This behavior pattern uses i-line and k-line.
(4) Station A entrance / stay (12: 31: 14-12: 33) → α boarding (12: 33-12: 45) → B station stay / exit (12: 45-12: 53: 01)
This behavior pattern uses the j-line.
(5) Station A entrance / stay (12: 31: 14-12: 40) → β boarding (12: 40-12: 52) → B station stay / exit (12: 52-12: 53: 01)
This behavior pattern uses the j-line.

When the used route is not uniquely determined and the calculated number of behavior patterns of the user is greater than a predetermined number, the certainty of each behavior pattern is low. In such a case, the probability of the user getting on the train of each route and the probability of the user being in the ticket gate of each station are obtained based on the conventional statistical information on the macro railway use, that is, the railway use index table 6T. It is reasonable.
In addition, the predetermined number compared with the number of user's action patterns may be described in the program 1 as a threshold value, and the calculation apparatus 3 may receive from a user. When the calculation device 3 receives a threshold value to be compared with the number of behavior patterns from the user, the CPU 31 displays an input screen (not shown) for that purpose on the display unit 38, and the user can input the threshold value from the input screen via the operation unit 39. Enter.

  On the other hand, if the route used is not uniquely determined and the number of calculated user behavior patterns is less than the predetermined number, the probability of evenly allocated according to the number of behavior patterns is set in the behavior pattern. Can be considered. For example, in the case of the above five patterns, a probability of 20% is assigned to each action pattern. However, the number of ticket gate logs registered in the ticket gate log table 1T is generally enormous, and the ticket gate log table 1T includes ticket gate logs similar to the ticket gate logs to be processed. Hereinafter, a ticket gate log similar to the ticket gate log to be processed is referred to as a similar log. The calculation device 3 performs probability calculation, which will be described later, using the similar log.

By collecting not only the ticket gate logs to be processed but also collecting similar logs and associating them with the ticket gate logs to be processed, the population of the ticket gates to be processed will increase and the certainty of the generated action pattern can be improved. it can.
As a concept, the ticket gate log can be regarded as a measurement result of a very rough sampling of passenger behavior. Since the sampling is rough, the behavior of the target user is not well understood, but there is no contradiction between the user behavior and the ticket gate log. In general, even if the sampling is rough, if the sampling timing is gradually shifted with respect to the same object and measurement is performed many times, the measurement object is accurately observed. Based on this concept, the same behavior pattern (flight / route) is measured (ticket gates), and a large number of data (ticket gate logs) with different sampling timings (ticket gate passage times) are collected little by little to estimate user behavior. The result will be more accurate.

The similar log considered first is another ticket gate log in which the user, the entry station, and the entry station are the same as the ticket gate log to be processed. For example, the behavior patterns of commuters, commuters, etc. are usually the same every day. The ticket gate log accumulated every day corresponding to such behavior patterns of commuters, commuters, etc. is the first similar log.
When considering commuting, attending school, and daily travel, the same user is likely to use the same flight. In addition, even if the same user does not use the same flight, there is a high possibility of using the same route.

  A second possible similar log is another ticket gate log in which the user and the participating station are the same as the ticket gate log to be processed. The entrance stations corresponding to the ticket gate log to be processed and the second similar log are different. The route corresponding to the second similar log may be different from the route corresponding to the ticket gate log to be processed. However, there is a high possibility that the route used in the vicinity of the participating station is common. For example, when a user departs for some reason from work or school, the entry station may be the same station as the daily action pattern even if the entrance station is a different station from the daily action pattern. High nature.

A third possible similar log is another ticket gate log whose entry station is the same as the ticket gate log to be processed and whose entry time is close to the entry time of the ticket gate log to be processed. The adjoining entry time here is a time having a difference within 1 minute before and after, for example. The third similar log is not similar data based on regularity in the user's railway use like the first and second similar logs. However, the behavior pattern based on the third similarity log is considered to be similar to the behavior pattern based on the ticket gate log to be processed from the proximity of the participation time and the user's behavior near the participation station.
Even if the users are different, the flight at the time of disembarkation related to the third similar log was the same as the flight of the action pattern based on the ticket gate log to be processed by attaching the condition that the participation time is close The possibility increases.

  It should be noted that other methods can be considered as a method for determining the similarity between the ticket gate log to be processed and another ticket gate log. For example, elements such as the season, month, day of the week, time zone, user attribute, etc. related to the behavior pattern of the user using the railway may be incorporated into the similarity determination.

The calculation device 3 extracts a ticket gate log similar to the ticket gate log to be processed from the ticket gate log table 1T according to the similarity determination criterion.
Hereinafter, an example of processing for extracting a similar log will be specifically described.
The calculation device 3 extracts, from the ticket gate log table 1T, ticket gate logs that have the same user, the same entry station and the same departure station, and a difference in date and time within three months with respect to the ticket gate log to be processed.
The calculation device 3 extracts, from the ticket gate log table 1T, ticket gate logs that have the same user and share the same station as the processing target ticket gate logs and that have a date and time difference within three months.
The calculation device 3 extracts from the ticket gate log table 1T the ticket gate logs that have different users, have the same entry station, and have a participation time within 1 minute before and after the ticket gate log to be processed.

The calculation device 3 gives a weight to the ticket gate log to be processed and the extracted similar log.
The similarity between the ticket gate log to be processed and the similar log decreases in the order of the first, second, and third similar logs. A similar log with higher similarity is considered to be closer to an action pattern generated based on a ticket gate log to be processed. Therefore, the calculation device 3 gives a greater weight to the similarity log with higher similarity. Then, the calculation device 3 gives the maximum weight to the ticket gate log to be processed.

Hereinafter, an example of giving specific weights to the ticket gate log and the similar log to be processed will be described.
A weight of 100 is assigned to the ticket gate log to be processed.
The weight given to the similar log is determined according to the similarity between the action pattern based on the similar log and the action pattern based on the ticket gate log to be processed.
For similar logs that correspond to behavior patterns with the same user and the same day of the week, if the usage date is within one week before and after the usage date of the ticket gate log to be processed, it is 80, within two weeks before and after 78, if it is within 3 weeks before and after, give a weight of 76. Thereafter, the weight given by 2 is reduced every time the usage date is further separated by one week.
For similar logs that correspond to behavior patterns with the same user and different days of the week, the usage date of the ticket gate log to be processed is 60 if the usage date is within 2 weeks, and 55 if the usage date is within 4 weeks. In the case of within 6 weeks before and after, a weight of 50 is given. Thereafter, the weight to be given by 5 is reduced every use week every two weeks.
For similar logs corresponding to behavior patterns with the same user and different AM / PM at the time, the weight obtained by multiplying the weight by 0.5 is used as the weight.
For similar logs corresponding to behavior patterns with the same user and different entrance stations, the weight obtained by multiplying the weight by 0.5 is used as the weight.
60 weights are uniformly assigned to similar logs corresponding to behavior patterns with different users.

The calculation device 3 collates the extracted similar log with the operation record table 3T and generates an action pattern based on the similar log. And the calculation apparatus 3 calculates a probability with respect to the action pattern based on the ticket gate log of a process target, and the action pattern based on a similar log.
First, 100% probability is equally divided by the number of generated behavior patterns. For example, when there are two action patterns based on the ticket gate log to be processed, a probability of 50% is assigned to each. Further, when there are three action patterns based on the extracted similar logs, a probability of 33.3% is assigned to each. Hereinafter, a probability obtained by equally dividing the probability of 100% by the number of behavior patterns is referred to as a base probability. In other words, the base probability is the reciprocal of the number of generated behavior patterns.

Next, the calculation device 3 associates the behavior pattern based on the ticket gate log to be processed with the behavior pattern based on the similarity log. The association here is to incorporate the weight and base probability in the action pattern based on the similar log into the probability of the ticket gate log to be processed.
When the behavior pattern based on the ticket gate log to be processed and the behavior pattern based on the similar log completely match, the calculation device 3 performs the association as it is. When the behavior pattern based on the ticket gate log to be processed and the behavior pattern based on the similarity log do not completely match, the calculation device 3 classifies the association based on the similarity of the behavior patterns. Here, various criteria for determining the similarity of behavior patterns may be set.

The following shows an example of association of action patterns when an action pattern based on a ticket gate log to be processed and an action pattern based on a similar log do not completely match.
In the behavior pattern based on the similarity log, the calculation device 3 does not associate if the last used flight does not match the last used flight of any behavior pattern based on the ticket gate log to be processed. This is because in such a case, the similarity in the behavior pattern is considered to be extremely low. In addition, it can be said that the process which is not matched is also a correction process in probability calculation.

  On the other hand, when the last usage flight of the behavior pattern based on the similarity log matches the last usage flight of any behavior pattern based on the ticket gate log to be processed, the calculation device 3 executes the association. In addition, the calculation device 3 uses the behavior pattern based on the similar log when the route of the last use flight matches even if the last use flight does not match the last use flight of any action pattern based on the ticket gate log to be processed. The association is executed. However, even when the calculation device 3 is associated, when the last used flight or the route of the last used flight in the action pattern based on the similar log matches a plurality of action patterns based on the ticket gate log to be processed, Execute the process. That is, the calculation device 3 divides the base probability of the action pattern based on the similar log by the number of action patterns based on the matching ticket gate log to be processed, and associates the divided base probability with the corresponding action pattern.

When the association is completed, the calculation device 3 calculates the probability for each action pattern based on the ticket gate log to be processed using the following equation (1).
Probability related to limited log = Σ (weight × base probability) / Σ (weight) (1)
However, the calculation device 3 excludes the weight of the action pattern based on the similar log that is not associated from the denominator weight of the equation (1).

FIG. 25 is an explanatory diagram showing a list of probability calculation elements for each behavior pattern generated based on the limited log.
The identifier is a number that identifies each behavior pattern. The attributes of the ticket gate log are the card ID, date / time, and weight of the ticket gate log. The base probability is a base probability equally divided for the ticket gate log and the similar log to be processed. The behavior pattern is a behavior pattern of the user generated based on the ticket gate log and the similar log to be processed.

The ticket gate logs corresponding to the identifiers [1] to [5] are the ticket gate logs to be processed.
The ticket gate logs corresponding to the identifiers [6] to [11] are examples of similar logs. In the example of FIG. 25, ID001, a log similar to March 25, ID001, a log similar to March 23, and ID003, a total of three logs similar to April 1, are extracted. If the card IDs are the same, the users are considered to be the same, so the user corresponding to the similar log on March 25 of the first case and March 23 of the second case will be included in the ticket gate log to be processed. It is the same as the corresponding user. On the other hand, the user corresponding to the third similar log on April 1 is different from the user corresponding to the ticket gate log to be processed.

The similar log in FIG. 25 will be described in the light of the condition example for extracting the above-mentioned similar log.
The first similar log corresponding to the identifiers [6] to [8] corresponds to the case where the users are the same, the entrance station and the exit station are common, and the date and time difference is within 3 months.
The second similar log corresponding to the identifier [9] corresponds to the case where the users are the same and the participating stations are common, and the date and time difference is within three months.
The third similar log corresponding to the identifiers [10] to [11] corresponds to a case where the users are different, the participating stations are common, and the participating time is within one minute before and after.

The weights to be given to the ticket gate log and the similar log to be processed in FIG.
A weight of 100 is given to the ticket gate log to be processed.
The first similar log corresponding to the identifiers [6] to [8] is a similar log when the user corresponds to the same action pattern on the day of the week and the use date is within one week before and after. As shown in FIG.
The second similar log corresponding to the identifier [10] corresponds to a behavior pattern in which the user is the same and the day of the week is different, and the entry station is different from the weight 60 when the usage date is within two weeks before and after. A weight of 30 multiplied by 0.5 is given.
The third similar log corresponding to the identifiers [10] to [11] is uniformly assigned a weight of 60 corresponding to different behavior patterns of users.

The action patterns of the identifiers [1] to [5] are five action patterns generated by comparing the ticket gate log in FIG. 21 with the shortest travel time in FIG. 22, the operation record in FIG. 23, and the route map in FIG. is there. Here, it is assumed that the ticket gate log to be processed is the limited log in FIG. The five behavior patterns generated based on the limited log in FIG. 21 are hereinafter referred to as original data.
Since the number of the original data is five, the base probability for the original data is 20%, with 100% probability being equally allocated.

It is assumed that the behavior patterns of the identifiers [6] to [11] are generated by comparing the extracted similar logs with operation records and the like.
Since the number of action patterns corresponding to the identifiers [6] to [8] is three, the base probability for these action patterns is equally allocated 100%, and each is 33.3%.
Since the number of action patterns corresponding to the identifier [9] is one, the base probability for this action pattern is 100%.
Since the number of action patterns corresponding to the identifiers [10] to [11] is two, the base probability for these action patterns is equally allocated 100%, and each is 50%.

The association of action patterns based on the similar log in FIG. 25 to the original data will be described.
The behavior patterns of the identifiers [6], [7], and [8] completely match the behavior patterns in the original data of the identifiers [1], [2], and [4], respectively. Accordingly, the calculation device 3 directly associates the behavior patterns of the identifiers [6], [7], and [8] with the original data of the identifiers [1], [2], and [4].
The behavior pattern of the identifier [9] does not completely match all the behavior patterns in the original data. However, the route of the last used flight in the action pattern of the identifier [9] is the i line, which matches the i line of the last used flight in the original data of the identifiers [1] and [2]. In this case, since the number of original data that can be associated is two [1] and [2], the calculation device 3 divides the base probability 100% corresponding to the behavior pattern of the identifier [9] by 2. 50% is allocated and associated with the original data of the identifiers [1] and [2].

The behavior pattern of the identifier [10] matches the Y flight on the i line that is the last used flight with the Y flight on the i line that is the last used flight of the original data of the identifier [2]. Accordingly, the calculation device 3 associates the behavior pattern with the identifier [10] with the original data with the identifier [2].
The behavior pattern of the identifier [11] does not completely match all the behavior patterns in the original data. Further, the zz flight of the last use flight in the behavior pattern of the identifier [11] is not included in all the original data. Furthermore, the l line of the last used flight in the behavior pattern of the identifier [11] is not included in all the original data. Therefore, the calculation device 3 does not associate the behavior pattern of the identifier [11] with any original data.
The weight of the similar log corresponding to the behavior patterns of the identifiers [10] and [11] is 60. However, the weight corresponding to the identifier [11] is excluded when calculating the denominator of the equation (1). The calculation device 3 sets the weight in the denominator to 30 which is half of 60 in order to use the behavior pattern of the identifier [11] as the denominator considering only the behavior pattern of the identifier [10] without associating the behavior pattern with the identifier [11].

Based on the above, when the expression (1) is executed, the probabilities related to the original data [1] to [5] are as follows.
[1] (100 × 20 + 80 × 33.3 + 30 × 50 + 60 × 0) / (100 + 80 + 30 + 30) ≈25.68%
[2] (100 × 20 + 80 × 33.3 + 30 × 50 + 60 × 50) / (100 + 80 + 30 + 30) ≈38.18%
[3] (100 × 20 + 80 × 0 + 30 × 0 + 60 × 0) / (100 + 80 + 30 + 30) ≈8.33%
[4] (100 × 20 + 80 × 33.3 + 30 × 0 + 60 × 0) / (100 + 80 + 30 + 30) ≈19.43%
[5] (100 × 20 + 80 × 0 + 30 × 0 + 60 × 0) / (100 + 80 + 30 + 30) ≈8.33%
In the example of FIG. 25, it is indicated that the behavior pattern of the identifier [2] has the highest probability of using the railway.

  26, 27, 28, and 29 are explanatory diagrams for explaining the non-limited log. FIG. 26 is an explanatory diagram showing a record example of the ticket gate log table 1T corresponding to the limited log. FIG. 27 is an explanatory diagram of a record example of the shortest movement time table 2T. FIG. 28 is an explanatory diagram of a record example of the operation record table 3T. In FIG. 28, the station arrival / departure time train on the way is omitted. FIG. 29 is an explanatory diagram showing an example of a route map. FIG. 29 is a route map visualizing the information in the route table 4T.

  From FIG. 26, it can be seen that the user entered the A station and entered the B station. From FIG. 29, there are five routes connecting the A station and the B station: i-line, j-line, k-line, i-line → k-line, and k-line → i-line. Does not log. From FIG. 27, the shortest movement time between A station and B station is 9-12 minutes. On the other hand, from FIG. 26, the elapsed time between station A and station B is 71 minutes and 12 seconds, which is considerably longer than the shortest travel time. From the operation record of FIG. 28, since the train operation interval is about several minutes, the number of user action patterns that can be generated from the ticket gate log of FIG. 26 is very large. When generation of an action pattern is started based on the ticket gate log of FIG. 26, for example, when the threshold value of the number of action patterns is 20, the number of action patterns based on the ticket gate log of FIG. 26 exceeds the threshold value. Therefore, the calculation device 3 classifies the ticket gate log in FIG. 26 as an unrestricted log.

The calculation device 3 accumulates non-limited logs in the RAM 32 in the process of looping many ticket gate logs. The calculation device 3 classifies and accumulates the accumulated non-limited logs according to the sections of the entrance station and the entry station. The calculation device 3 distributes the number of limited logs classified and tabulated, that is, the number of users by the ratio of the number of users of each route and station based on the railway usage index table 6T, that is, the probability.
In the railway usage index table 6T, since the number of railway users obtained macroscopically from visual observation by a station employee or the like, camera observation, etc. is registered, the processing of the calculation device 3 based on the unrestricted log follows a macro approach. The number of train users will be calculated.

When the railway usage index table 6T stores the number of railway users for each time zone, the calculation device 3 corresponds to the number of limited logs that are classified and aggregated to the processing target date and time received by the reception unit 302. You may distribute by the ratio of the number of railway users in a time zone.
Moreover, when the route handled by the railway user number calculation system is divided into a plurality of route areas, the calculation device 3 may distribute a non-restricted log for each route area. The calculation device 3 counts the number of non-limited logs by route area, and distributes the total number of non-limited logs, that is, the number of users based on the ratio of the number of users of each route and station based on the railway usage index table 6T. .
The above is the action pattern generation process and the probability calculation process for the confirmation log, the route confirmation log, the limited log, and the non-limited log in the case classification process.

FIG. 30 is a flowchart illustrating an example of the procedure of the case classification process. FIG. 30 corresponds to step S104 in FIG.
The CPU 31 determines whether or not the ticket gate log to be processed is a route confirmation log (step S201). In step S201, the CPU 31 determines whether or not there is one route used by the user from the entrance and exit stations corresponding to the ticket gate log and the route table 4T. If the CPU 31 determines that the ticket gate log is a route confirmation log (step S201: YES), the CPU 31 executes route confirmation log processing (step S202) and returns.

  When the CPU 31 determines that the ticket gate log is not the route confirmation log (step S201: NO), the ticket gate log is compared with the shortest travel time table 2T, the operation record table 3T, and the route table 4T, and the ungenerated corresponding to the ticket gate log. One action pattern is generated (step S203). The CPU 31 increments the counter variable n (Step S204). The counter variable n corresponds to the number of generated behavior patterns. CPU31 determines whether n is larger than a threshold value (step S205). If the CPU 31 determines that n is greater than the threshold (step S205: YES), the CPU 31 executes non-limiting log processing (step S206) and returns.

  When determining that n is not larger than the threshold (step S205: NO), the CPU 31 determines whether another behavior pattern can be generated (step S207). CPU31 returns a process to step S203, when it determines with another action pattern being generable (step S207: YES). When determining that another action pattern cannot be generated (step S207: NO), the CPU 31 determines whether n is 1 (step S208).

  If the CPU 31 determines that n is 1 (step S208: YES), the CPU 31 executes a confirmation log process (step S209) and returns. If the CPU 31 determines that n is not 1 (step S208: NO), it executes a limited log process (step S210) and returns.

  The confirmation log process is a process of setting 1 as a probability of a calculation result for one generated action pattern. Further, in the confirmation log process, the CPU 31 sets 1 as the probability of the calculation result even in the case of an action pattern that does not get on the train from the station where it entered and participates as it is.

  The threshold value in step S205 is a boundary value for distributing the ticket gate log to be processed to an unrestricted log. When the threshold value is set to a small value, the number of non-limited logs increases and the number of confirmed logs and limited logs decreases. On the other hand, when the threshold value is set to a large value, the number of non-limited logs decreases, and the number of confirmed logs and limited logs increases. Calculation of the number of railway users based on unrestricted logs depends on conventional macro statistical data. The calculation of the number of railway users based on the fixed log and the limited log depends on an approach for tracking micro individual behavior. The calculation device 3 calculates the number of railway users by hybrid processing using both macro and micro processing, but the ratio of macro and micro dependence can be adjusted by the threshold value in step S205. For example, when the threshold value = 0 is set, the calculation device 3 can also calculate the number of railway users for all the ticket gate logs from the conventional macro processing.

FIG. 31 is a flowchart illustrating an exemplary procedure of route determination log processing.
The CPU 31 calculates the elapsed time from the entry time and the entry time of the ticket gate log to be processed (step S301). The CPU 31 searches the shortest travel time table 2T for the shortest travel time between the entrance station and the exit station in the ticket gate log (step S302). CPU31 determines whether the difference of elapsed time and the shortest movement time is longer than a threshold value (step S303).

In step S303, the CPU 31 stores the determination result in the work flag. In later steps S307 and S310, the CPU 31 refers to the work flag to determine whether the behavior pattern is a pattern that stays at the station for a predetermined time or more and a pattern that does not stay at the station. .
Also, staying at the station here means that the user stays in the ticket gate of the station.

  When determining that the difference between the elapsed time and the shortest movement time is longer than the threshold (step S303: YES), the CPU 31 generates one ungenerated action pattern that stays at one station for a predetermined time or longer (step S304). . In step S304, the CPU 31 ignores the case where the user stays at a plurality of stations for a predetermined time or more as a rare case. If the CPU 31 determines that the difference between the elapsed time and the shortest travel time is not longer than the threshold value (step S303: NO), the CPU 31 generates one ungenerated action pattern that does not stay in the station for a predetermined time (step S305).

  The CPU 31 increments the counter variable l (step S306). The counter variable l corresponds to the number of action patterns generated. The CPU 31 refers to the work flag set in step S303, and determines whether or not another action pattern can be generated for either the station stay or station non-stay pattern (step S307). When determining that another behavior pattern can be generated (step S307: YES), the CPU 31 returns the process to step S304 or step S305 according to the work flag set in step S303. That is, when the work flag is a station stay pattern, the CPU 31 returns the process to step S304. CPU31 returns a process to step S305, when a work flag is a pattern which is not a station stay.

If the CPU 31 determines that another action pattern cannot be generated (step S307: NO), it determines whether l is 1 (step S308). If the CPU 31 determines that l is 1 (step S308: YES), it executes a confirmation log process (step S309) and returns. When determining that l is not 1 (step S308: NO), the CPU 31 refers to the working flag set in step S303 and determines whether or not the station stay pattern is set (step S310). If the CPU 31 determines that the pattern is a station stay pattern (step S310: YES), the CPU 31 refers to the station facility scale factor table 5T and calculates the probability of considering the station facilities for each generated action pattern (step S311). To return. When it is determined that the pattern is not a station stay pattern (step S310: NO), the CPU 31 executes a similar log process (step S312) and returns.
The similar log process is a probability calculation process using a similar log corresponding to a ticket gate log to be processed. Similar log processing is the same as limited log processing. The limited log process will be described later.

FIG. 32 is a flowchart illustrating an example of the procedure of non-limiting log processing. The non-limiting log process broadly includes a process performed in step S206 in FIG. 30 and a part of the process for counting the number of users performed in step S106 in FIG.
In the non-limited log process, the number of non-limited logs is totaled. However, the total number of unrestricted logs is not determined until after the loop processing from step S103 to step S105 in FIG. FIG. 32 is an example of a preparation process procedure for counting the number of non-limited logs.
The CPU 31 increments the counter variable m (step S401) and returns. The counter variable m corresponds to the number of non-limited logs.
When the loop processing from step S103 to step S105 in FIG. 11 is completed, the counter variable m stores the number of unrestricted logs after aggregation.

FIG. 33 is a flowchart illustrating an example of a totaling process procedure based on a non-limited log. The procedure of FIG. 33 is a part of the procedure of the aggregation process in step S106 in FIG.
The CPU 31 determines the value of the counter variable m as the total number of unrestricted logs (step S402). The CPU 31 refers to the railway usage index table 6T, assigns the total number of non-limited logs to the number of users of each route and each station (step S403), and ends a part of the total processing.

In step S401 of FIG. 32 included in the loop for processing the extracted ticket gate logs one by one, the railway usage index table 6T is referred to, and the number of users on each route and each station in each non-limited log to be processed. A ratio, ie probability, may be set. In such a case, the calculation device 3 adds up the number of users in step S106 of FIG. 11 based on the probabilities set for each non-limited log in step S401 of FIG.
However, the number of users based on the non-restricted log is distributed to each route and each station on the basis of macroscopic statistical data. Therefore, as shown in FIGS. And the processing efficiency is higher if it is distributed to each station.

FIG. 34 is a flowchart illustrating an example of the procedure of limited log processing. FIG. 34 corresponds to step S210 in FIG. FIG. 34 also corresponds to step S312 of FIG. When FIG. 34 corresponds to step S312 of FIG. 31, the limited log may be replaced with a route determination log.
The CPU 31 executes a similar log extraction process (step S501). The similar log extraction process is a process of extracting a ticket gate log similar to the ticket gate log to be processed from the ticket gate log table 1T. The CPU 31 executes a weighting process (step S502). The weighting process is a process of giving a weight according to the degree of coincidence or similarity with the limited log to the limited log and the similar log. Details of the similar log extraction process and the weighting process will be described later.

The CPU 31 collates the similar log with the shortest travel time table 2T, the operation record table 3T, and the route table 4T, and generates an action pattern based on the similar log (step S503).
Note that the generation of the behavior pattern corresponding to the original data is completed in the loop processing from step S203 to step S207 in FIG. The original data is an action pattern generated based on the ticket gate log to be processed.
When FIG. 34 is made to correspond to step S312 of FIG. 31, the generation of the action pattern corresponding to the original data is completed by the loop processing from step S304 to step S307 of FIG.

  The CPU 31 calculates a base probability (step S504). In step S <b> 504, the CPU 31 calculates the probability of being evenly allocated by the number of original data. In step S <b> 504, the CPU 31 also calculates the probability of evenly allocating the number of action patterns generated based on the similar log.

The CPU 31 compares the action pattern of the original data with the action pattern of the similar log, and executes association (step S505). In step S505, when the behavior pattern of the original data and the behavior pattern of the similar log completely match, the CPU 31 associates the behavior patterns that completely match with each other.
On the other hand, in step S505, when the route of the last use flight or the last use flight of the action pattern based on the similar log is not included in the last use flight or the last use flight route of the original data, the CPU 31 is based on the similar log. It is determined that there is no possibility that the user has executed the behavior pattern. Then, the CPU 31 does not associate an action pattern based on the similarity log with the original data.
In step S505, when the last used flight or the route of the last used flight based on the action pattern based on the similar log is included in the last used flight or the last used flight route of the original data, the CPU 31 determines the action pattern based on the similar log. Is associated with the original data. However, when there are a plurality of pieces of original data that can be associated, the CPU 31 equally allocates the base probabilities corresponding to the action patterns based on the similar log being processed to the plurality of pieces of original data.

  CPU31 performs the probability calculation which concerns on a limitation log based on (1) Formula (step S506). The CPU 31 sets the calculated probability as the behavior pattern of the original data (step S507) and returns.

FIG. 35 is a flowchart illustrating an example of a procedure of similar log extraction processing.
The CPU 31 extracts, from the ticket gate log table 1T, a record in which the card ID, the entry station, and the entry station match each other and the difference in use date and time is within three months for the ticket gate log to be processed (step S601). . Here, the match of the card ID means the match of the user. The CPU 31 extracts, from the ticket gate log table 1T, a record in which the card ID and the participating station match and the use date difference is within three months for the ticket gate log to be processed (step S602). However, in step S602, the CPU 31 excludes the record extracted in step S601. The CPU 31 extracts records in which the card ID is different, the entry stations match, and the entry time difference is within one minute from the ticket gate log to be processed (step S603). CPU31 returns, when the process of step S603 is complete | finished.

  Note that there are other similar log determination methods in the similar log extraction process of FIG. For example, in step S602, the match of the entry station is set as one of the determination conditions, but the match of the entry station may be replaced with the match of the entry station. In step S603, one of the determination conditions is that the difference in the entry time is within 1 minute, but the difference in the entry time is within 1 minute, for example, that the difference in the entry time is within 5 minutes. May be substituted. In FIG. 35, three types of similar log records are extracted from the ticket gate log table 1T according to the degree of similarity with the ticket gate log to be processed, but one type, two types, or four or more types of similar log records are extracted. Records may be extracted from the ticket gate log table 1T.

FIG. 36 is a flowchart illustrating an example of a weighting process.
The CPU 31 assigns a weight of 100 to the ticket gate log to be processed (step S701). The weighting process after step S702 is a process for giving a weight to the similar log.
The CPU 31 selects one unprocessed similar log from the similar logs extracted by the similar log extraction process of FIG. 35 (step S702). The CPU 31 determines whether or not the card ID of the selected similar log is the same as the card ID of the ticket gate log to be processed (step S703). Here, the match of the card ID means the match of the user.

  If the CPU 31 determines that the card ID of the selected similar log is not the same as the card ID of the ticket gate log to be processed (step S703: NO), the CPU 31 gives a weight of 60 to the similar log to be processed (step S704). The process proceeds to step S712. That is, the CPU 31 uniformly assigns a weight of 60 to similar logs with the same usage date but different users for the ticket gate log to be processed. In the similar log extraction process of FIG. 35, when the users are different, similar logs with a difference in entry time within one minute are extracted, so that all similar logs with different users have the same usage date.

  If the CPU 31 determines that the card ID of the selected similar log is the same as the card ID of the ticket gate log to be processed (step S703: YES), the day of use of the similar log is the date of use of the ticket gate log to be processed. It is determined whether it is the same as the day of the week (step S705). When the CPU 31 determines that the day of use of the similar log is the same as the day of use of the ticket gate log to be processed (step S705: YES), the CPU 31 assigns a maximum weight of 80 to the similar log to be processed (step S705). Step S706). In step S706, the CPU 31 determines that the usage date of the ticket gate log to be processed is 80 for a similar log whose usage date is within one week before and after, 78 for a similar log whose usage date is within two weeks before and after, and a usage date of A weight of 76 is assigned to similar logs within three weeks before and after. As described above, in step S706, the CPU 31 linearly assigns a maximum weight of 80 to the similar log to be processed.

  If the CPU 31 determines that the day of use of the similar log is not the same as the day of use of the ticket gate log to be processed (step S705: NO), the CPU 31 assigns a maximum weight of 60 to the similar log to be processed (step S705). S707). In step S <b> 707, the CPU 31 compares the usage date of the processing target ticket gate log with respect to the usage date of the processing target ticket gate log to 60 for a similar log whose usage date is within 2 weeks before and after, and the usage date within 4 weeks before and after. A weight of 55 is assigned to the similar log, and a weight of 50 is assigned to the similar log whose use date is within 6 weeks. As described above, in step S707, the CPU 31 linearly assigns a maximum of 60 weights to the similar log to be processed.

  The CPU 31 determines whether or not the AM / PM of the usage date / time of the similar log is the same as the AM / PM of the usage date / time of the ticket gate log to be processed (step S708). If the CPU 31 determines that the AM / PM of the usage date / time of the similar log is the same as the AM / PM of the usage date / time of the ticket gate log to be processed (step S708: YES), the CPU 31 advances the processing to step S710. If the CPU 31 determines that the AM / PM of the usage date / time of the similar log is not the same as the AM / PM of the usage date / time of the ticket gate log to be processed (step S708: NO), the CPU 31 assigns 0 to the weight assigned to the processing target similar log. .5 (step S709), and the process proceeds to step S710.

  The CPU 31 determines whether or not the entrance station of the similar log is the same as the entrance station of the ticket gate log to be processed (step S710). When the CPU 31 determines that the entrance station of the similar log is the same as the entrance station of the ticket gate log to be processed (step S710: YES), the CPU 31 proceeds with the process to step S712. When the CPU 31 determines that the entrance station of the similar log is not the same as the entrance station of the ticket gate log to be processed (step S710: NO), the CPU 31 multiplies the weight given to the similar log to be processed by 0.5 (step S711). Then, the process proceeds to step S712.

  The CPU 31 determines whether there is an unprocessed similar log for the similar logs extracted by the similar log extraction process (step S712). If the CPU 31 determines that there is an unprocessed similar log extracted by the similar log extraction process (step S712: YES), the CPU 31 returns the process to step S702. If the CPU 31 determines that there is no unprocessed similar log extracted by the similar log extraction process (step S712: NO), the CPU 31 returns.

Note that there may be other weighting methods in the weighting process of FIG. For example, the weighting process may include a determination process for determining whether the route of the last used flight related to the ticket gate log to be processed is the same as the route of the last used flight related to the similar log. When the route of the last use flight is the same, the weight given to the similar log is made heavier than when the route of the last use flight is different.
Further, in step S704 when the users are different, a branching process for assigning a weight other than 60 may be added according to the similarity of the ticket gate log or the similarity of the action pattern based on the ticket gate log.

  When calculation of the probability corresponding to each action pattern generated based on the ticket gate log extracted from the ticket gate log table 1T at the processing target date and time is completed, the location of the user at the flight or station at the time of the processing target date and time is probabilistic. It will be expressed in. Since one ticket gate log corresponds to one user, the probability calculated for the action pattern based on one ticket gate is the expected value of one user as it is. The calculation device 3 calculates how many users were in each flight and each station by counting the probabilities calculated for each behavior pattern of each user. That is, the calculation device 3 analyzes the use situation of the railway.

  Here, regarding the totaling method, two methods can be considered. One counting method is a method of using a plurality of behavior patterns generated for one user and their probabilities as they are. For example, when two action patterns and probabilities of 70% for the A action pattern and 30% for the B action pattern are respectively obtained for one ticket gate log, 0.7 persons are assigned to the A action pattern, and the B action pattern Set 0.3. When the action pattern and the probability are determined, the location of the user at the processing date and time is determined uniquely and probabilistically. Therefore, 0.7 person at the position of the A action pattern from one ticket gate log, the B action pattern Add 0.3 to each position. By repeating the same calculation for all the extracted ticket gate logs, the number of railway users at the processing date and time is calculated as a snapshot.

  Another tabulation method is a method in which the behavior pattern having the highest calculated probability is represented as a unique behavior pattern. When this counting method is applied to the above example, one person is assigned to the A action pattern having the highest probability for one ticket gate log. The same process is repeated for all ticket gate logs. Then, by counting the allocated number of users for each behavior pattern, the number of railway users at the processing target date and time is calculated as a snapshot.

  The calculation device 3 may count the number of railway users by combining the above-described two counting methods. For example, the calculation device 3 determines a threshold value corresponding to the probability of the behavior pattern, and uses two types of aggregation methods using the threshold value as a boundary. More specifically, when the probability of a certain behavior pattern is, for example, 80% or more among the plurality of generated behavior patterns, the calculation device 3 sets the behavior pattern as the only behavior pattern. On the other hand, when all of the plurality of generated behavior patterns are less than 80%, the calculation device 3 totals the probabilities for each behavior pattern.

  Note that the number of railway users based on the unrestricted log is allocated to each flight and each station according to the procedure of FIG. 33 by applying macro statistical data. Therefore, the calculation device 3 adds the number of railway users calculated macroscopically based on the non-restricted log to the number of railway users tabulated from the above-described microscopically generated behavior pattern probabilities.

According to the calculation device 3, the number of railway users can be accurately calculated based on the ticket gate log.
The calculation device 3 classifies the ticket gate log into four types: a confirmed log, a route confirmed log, a limited log, and a non-limited log. The certainty of the user's action pattern generated based on the ticket gate log decreases in the order of the confirmed log, route confirmed log, limited log, and non-limited log. One action pattern is determined from a confirmed log with a high degree of certainty. In general, a plurality of behavior patterns are generated from the route confirmation log, and the probability corresponding to each behavior pattern is calculated based on the station equipment. For the action pattern generated from the limited log with a certain degree of certainty, the probability is calculated using the similar log. From the unrestricted log with a low degree of certainty, the number of users or the probability according to macro statistical data is obtained. As described above, the calculation device 3 classifies the ticket gate logs according to the difference in certainty of the behavior pattern, and calculates the number of railway users by a wide range of calculation methods from micro to macro by properly using the classified ticket gate logs.

  The program 1P for operating the calculation device 3 that is a computer may be read from the optical disc 3a via the disc drive 34. Alternatively, the program 1P may be read from an external information processing device or recording device via the communication unit 36. Furthermore, a semiconductor memory 3c such as a flash memory in which the program 1P is recorded may be mounted in the calculation device 3.

  Regarding the above embodiment, the following additional notes are disclosed.

(Appendix 1)
In the calculation device that calculates the number of users of the railway based on the passage record of the ticket gates in the railway,
A storage unit that stores passage records including entry and departure times when multiple users pass through the ticket gates, train operation time information, route connection information, railway facility information, and statistical data on the number of users When,
Based on the passing record, operation time information, and route connection information stored in the storage unit, a predetermined time in the user's flight or the station ticket gate related to the passing record stored in the storage unit A generation unit that generates one or a plurality of behavior patterns related to the above stay;
Based on the facility information stored in the storage unit, the passage record or the statistical data of the number of users, a calculation unit that calculates the probability that the user has performed the action corresponding to the action pattern generated by the generation unit;
A calculation device comprising: a calculation unit that calculates the number of railway users according to the behavior pattern generated by the generation unit based on the probability calculated by the calculation unit.

(Appendix 2)
A reception unit for receiving the date and time;
The operation time information stored in the storage unit includes the operation time of each train by route,
The generation unit generates an action pattern of a user using a passage record in which an entry time is before the date and time accepted by the reception unit and an entry time is after the date and time,
The calculation unit is configured to determine a probability that a user has boarded a flight related to the behavior pattern generated by the generation unit or a probability of staying in a ticket gate of a station for a predetermined time or more at the date and time received by the reception unit. Calculate
The calculation unit calculates the number of users who have boarded a flight related to the behavior pattern generated by the generation unit or the number of users who stayed in the ticket gate at the date and time received by the generation unit. 1. The calculation device according to 1.

(Appendix 3)
The calculation device according to claim 2, wherein the generation unit generates a behavior pattern of a user based on a stay time in a ticket gate of a station, a train transfer between routes, a route or a flight.

(Appendix 4)
The train operation time information stored in the storage unit includes the shortest travel time in a predetermined section,
A first determination unit that determines whether or not there is one route that can be used in the user's behavior pattern, based on the passage record stored in the storage unit and the information on the connection relationship of the route;
Based on the passage record and operation time information stored in the storage unit, it is determined whether or not the time difference between the shortest travel time from the entry station to the entry station and the elapsed time from the entry time to the entry time is greater than or equal to a threshold value. A second determination unit, and
The generating unit generates an action pattern for staying in a ticket gate of a station for a predetermined time or more when the first determining unit determines that the pattern is one and the second determining unit determines that the threshold is equal to or greater than a threshold value. The calculation device according to any one of Notes 1 to 3.

(Appendix 5)
The information on the facilities in the railway stored in the storage unit includes information corresponding to the scale or contents of the facilities of the station,
The calculation unit according to any one of Supplementary Note 1 to Supplementary Note 4, wherein the calculation unit calculates a probability based on the information.

(Appendix 6)
If the number of action patterns generated by the generation unit is less than or equal to a predetermined number, the admission time is earlier than the date and time received by the reception unit, and the passing time is similar to the passing record in which the entry time is later than the date and time. An extraction unit for extracting a two-pass record from the storage unit;
A weight assigning unit that assigns a weight according to the degree of similarity to the passing record to the second passing record extracted by the extracting unit;
Based on the second passage record extracted by the extraction unit and the contents of the storage unit, a second generation unit that generates a user's behavior pattern related to the second passage record extracted by the extraction unit;
An association unit that associates the behavior patterns with each other based on the degree of coincidence or similarity between the behavior patterns generated by the generation unit and the second generation unit, respectively.
The calculation unit calculates a probability based on the weight given by the weight assigning unit, the behavior patterns associated by the association unit, and the number of behavior patterns generated by the generation unit and the second generation unit, respectively. The calculation device according to Supplementary Note 2 or Supplementary Note 3.

(Appendix 7)
The association unit associates the behavior patterns with each other when the departure station associated with the behavior pattern generated by the generation unit and the second generation unit and the flight of the train that has arrived at the departure station are the same. The calculation device described.

(Appendix 8)
The calculation device according to claim 6 or 7, wherein the calculation unit calculates a probability based on each of numerical values corresponding to the number of action patterns generated by the generation unit and the second generation unit, respectively.

(Appendix 9)
The calculation device according to any one of Supplementary Note 1 to Supplementary Note 8, wherein the calculation unit sets 1 as a probability of a calculation result when the behavior pattern generated by the generation unit is one.

(Appendix 10)
The statistical data of the number of users stored in the storage unit includes the average number of users who have been on the train of each route or the average number of users who have stayed in the ticket gates of each station,
When the number of behavior patterns generated by the generation unit exceeds a predetermined number, the calculation unit is based on statistical data on the number of users stored in the storage unit, and the user who has been on the train on each route The calculation device according to any one of appendix 1 to appendix 9, which calculates the number or the number of users staying in the ticket gates of each station.

(Appendix 11)
The train operation time information stored in the storage unit includes an operation record or a timetable. The calculation device according to any one of appendix 1 to appendix 9.

(Appendix 12)
In a program for causing a computer having a storage unit that stores a passage record of a ticket gate in a railway to execute a process of calculating the number of users of the railway,
The memory in which a passage record including entrance times and entry times when a plurality of users pass through the ticket gate, train operation time information, route connection information, railway facility information, and statistical data on the number of users is stored. Based on the passage record, operation time information related to the section, and information on the connection relation of the route, boarding the user's flight related to the passage record stored in the storage section or staying for a predetermined time or more in the ticket gate of the station Generate one or more such behavior patterns,
Based on the equipment information stored in the storage unit, the passage record or the statistical data of the number of users, calculate the probability that the user has performed the action corresponding to the generated action pattern,
A program that causes a computer to execute a process of calculating the number of railway users according to the generated behavior pattern based on the calculated probability.

(Appendix 13)
In a calculation method for calculating the number of railway users by a computer having a storage unit and a control unit in which passage records of ticket gates in a railway are stored,
The memory in which a passage record including entrance times and entry times when a plurality of users pass through the ticket gate, train operation time information, route connection information, railway facility information, and statistical data on the number of users is stored. Based on the passage record, operation time information related to the section, and information on the connection relation of the route, boarding the user's flight related to the passage record stored in the storage section or staying for a predetermined time or more in the ticket gate of the station One or more action patterns are generated by the control unit,
Based on the facility information stored in the storage unit, the passage record or the statistical data of the number of users, the probability that the user has performed the action corresponding to the generated action pattern is calculated by the control unit,
A calculation method for calculating the number of railway users according to the generated behavior pattern by the control unit based on the calculated probability.

(Appendix 14)
Based on the generated behavior pattern, the passing record is classified by the control unit,
The calculation method according to claim 13, wherein a probability is calculated by the control unit in accordance with the classified passage record.

3. Calculation device (computer)
31 CPU (control unit, generation unit, calculation unit, calculation unit, reception unit, first determination unit, second determination unit, extraction unit, weight assignment unit, second generation unit, association unit)
34 Hard disk (storage unit)
1P program 1T ticket gate log table 2T shortest travel time table 3T operation record table 4T route table 5T station equipment scale factor table 6T railway use index table

Claims (10)

In the calculation device that calculates the number of users of the railway based on the passage record of the ticket gates in the railway,
A storage unit that stores passage records including entry and departure times when multiple users pass through the ticket gates, train operation time information, route connection information, railway facility information, and statistical data on the number of users When,
Based on the passing record, operation time information, and route connection information stored in the storage unit, a predetermined time in the user's flight or the station ticket gate related to the passing record stored in the storage unit A generation unit that generates one or a plurality of behavior patterns related to the above stay;
Based on the facility information stored in the storage unit, the passage record or the statistical data of the number of users, a calculation unit that calculates the probability that the user has performed the action corresponding to the action pattern generated by the generation unit;
A calculation device comprising: a calculation unit that calculates the number of railway users according to the behavior pattern generated by the generation unit based on the probability calculated by the calculation unit. A reception unit for receiving the date and time;
The operation time information stored in the storage unit includes the operation time of each train by route,
The generation unit generates an action pattern of a user using a passage record in which an entry time is before the date and time accepted by the reception unit and an entry time is after the date and time,
The calculation unit is configured to determine a probability that a user has boarded a flight related to the behavior pattern generated by the generation unit or a probability of staying in a ticket gate of a station for a predetermined time or more at the date and time received by the reception unit. Calculate
The calculation unit calculates the number of users who have boarded a flight related to the behavior pattern generated by the generation unit or the number of users who stayed in a ticket gate at the date and time received by the reception unit. Item 2. The calculation device according to Item 1. The calculation device according to claim 2, wherein the generation unit generates a behavior pattern of a user based on a stay time in a ticket gate of a station, a train transfer between routes, a route or a flight. The train operation time information stored in the storage unit includes the shortest travel time in a predetermined section,
A first determination unit that determines whether or not there is one route that can be used in the user's behavior pattern, based on the passage record stored in the storage unit and the information on the connection relationship of the route;
Based on the passage record and operation time information stored in the storage unit, it is determined whether or not the time difference between the shortest travel time from the entry station to the entry station and the elapsed time from the entry time to the entry time is greater than or equal to a threshold value. A second determination unit, and
The generation unit generates an action pattern for staying in a ticket gate of a station for a predetermined time or more when the first determination unit determines that the pattern is one and the second determination unit determines that the threshold is equal to or greater than a threshold value. The calculation device according to any one of claims 1 to 3. The information on the facilities in the railway stored in the storage unit includes information corresponding to the scale or contents of the facilities of the station,
The calculation device according to any one of claims 1 to 4, wherein the calculation unit calculates a probability based on the information. If the number of action patterns generated by the generation unit is less than or equal to a predetermined number, the admission time is earlier than the date and time received by the reception unit, and the passing time is similar to the passing record in which the entry time is later than the date and time. An extraction unit for extracting a two-pass record from the storage unit;
A weight assigning unit that assigns a weight according to the degree of similarity to the passing record to the second passing record extracted by the extracting unit;
Based on the second passage record extracted by the extraction unit and the contents of the storage unit, a second generation unit that generates a user's behavior pattern related to the second passage record extracted by the extraction unit;
An association unit that associates the behavior patterns with each other based on the degree of coincidence or similarity between the behavior patterns generated by the generation unit and the second generation unit, respectively.
The calculation unit calculates a probability based on the weight given by the weight assigning unit, the behavior patterns associated by the association unit, and the number of behavior patterns generated by the generation unit and the second generation unit, respectively. The calculation device according to claim 2 or 3. The association unit associates the behavior patterns with each other when the departure station related to the behavior pattern generated by the generation unit and the second generation unit is the same as the flight of the train that has arrived at the departure station. The calculation device described in 1. The calculation device according to claim 6, wherein the calculation unit calculates a probability based on each of numerical values corresponding to the number of action patterns generated by the generation unit and the second generation unit, respectively. In a program for causing a computer having a storage unit that stores a passage record of a ticket gate in a railway to execute a process of calculating the number of users of the railway,
The memory in which a passage record including entrance times and entry times when a plurality of users pass through the ticket gate, train operation time information, route connection information, railway facility information, and statistical data on the number of users is stored. Based on the passage record, operation time information related to the section, and information on the connection relation of the route, boarding the user's flight related to the passage record stored in the storage section or staying for a predetermined time or more in the ticket gate of the station Generate one or more such behavior patterns,
Based on the equipment information stored in the storage unit, the passage record or the statistical data of the number of users, calculate the probability that the user has performed the action corresponding to the generated action pattern,
A program that causes a computer to execute a process of calculating the number of railway users according to the generated behavior pattern based on the calculated probability. In a calculation method for calculating the number of railway users by a computer having a storage unit and a control unit in which passage records of ticket gates in a railway are stored,
The memory in which a passage record including entrance times and entry times when a plurality of users pass through the ticket gate, train operation time information, route connection information, railway facility information, and statistical data on the number of users is stored. Based on the passage record, operation time information related to the section, and information on the connection relation of the route, boarding the user's flight related to the passage record stored in the storage section or staying for a predetermined time or more in the ticket gate of the station One or more action patterns are generated by the control unit,
Based on the facility information stored in the storage unit, the passage record or the statistical data of the number of users, the probability that the user has performed the action corresponding to the generated action pattern is calculated by the control unit,
A calculation method for calculating the number of railway users according to the generated behavior pattern by the control unit based on the calculated probability.

Images