JP2013109664A - Congestion prediction device, congestion prediction method, and congestion prediction program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform highly-accurate congestion prediction.SOLUTION: A congestion prediction device 300 includes a procedure information collection function 340 as an actual congestion level acquisition part to acquire an actual congestion level before a predetermined reference time for a congestion prediction, a daily congestion pattern control function 350 as an actual fluctuation pattern calculation part to calculate an actual fluctuation pattern showing fluctuation of the actual congestion level, and a corrected congestion pattern determination function 360 as a corrected fluctuation pattern determination part to compare the calculated actual fluctuation pattern with an event-base fluctuation pattern stored in a congestion pattern DB 240 as an event-base fluctuation pattern information DB before the reference time, thereby determining the event-base fluctuation pattern approximate to the actual fluctuation pattern.

Description

  The present invention relates to a congestion prediction device, a congestion prediction method, and a congestion prediction program that perform congestion prediction, and more particularly to a technology for performing congestion prediction of a store that provides services to customers.

Conventionally, it has been proposed to perform congestion prediction with higher accuracy by correcting a congestion prediction predicted in advance based on an actually occurring event.
For example, when forecasting congestion of any parking lot, when the amount of congestion in the surrounding parking lot is larger than the average, that is, when a positive difference from the average amount of congestion in the surrounding parking lot is recognized There has been a technique for correcting a large amount of congestion prediction in an arbitrary parking lot (see Patent Document 1).

JP 2009-2111253 (paragraphs 0043-0056)

  However, in the technique described in Patent Document 1, when the congestion prediction is inquired, the congestion prediction is corrected using information at only one time point that is the inquiry time point. In many cases, there was a large gap between the amount of congestion. For example, when a local event is held in the morning of a specific store in the morning, the amount of afternoon congestion may increase, or when the local event occurs in the afternoon, the amount of afternoon congestion may decrease. . For this reason, there is a problem that high-precision congestion prediction cannot be performed sufficiently.

  This invention is made | formed in view of the said problem, and makes it a subject to provide the congestion prediction apparatus, the congestion prediction method, and congestion prediction program which can perform highly accurate congestion prediction.

  The inventor of the present invention pays attention to the fact that the difference (variation) between the congestion prediction and the actual congestion amount occurs due to the influence of various events, and the difference between the past congestion prediction and the past actual congestion amount. (Fluctuation) was patternized by event (hereinafter sometimes referred to as “variation pattern by event”), and it was devised to correct the congestion prediction based on the event-specific variation pattern. Here, for example, the event that influences the weather condition of the area where the store exists, the characteristics of the area where the store exists, the occurrence of an event in the area where the store exists, the failure of the equipment installed in the store, etc. It is conceivable that the event-specific variation pattern can be calculated considering one or more of these events.

  That is, in order to solve the above-described problem, the congestion prediction device according to the present invention is a congestion prediction device that predicts the congestion status of a store as a congestion amount at each time, and the congestion prediction that is a prediction of the congestion status of the store is performed. At least one event-specific variation pattern that is a variation pattern for the congestion prediction information DB that is stored in advance and that is calculated in consideration of one or a plurality of events, and is stored in advance. An actual congestion amount acquisition unit that acquires an actual congestion amount before a predetermined reference time for the congestion prediction, and an actual variation pattern that indicates a change in the actual congestion amount. By comparing the calculated actual variation pattern and the event-specific variation pattern stored in the event-specific variation pattern information DB before the reference time, A corrected variation pattern determining unit that determines an event-specific variation pattern that approximates a pattern, and a portion after the reference time of the congestion prediction stored in the congestion prediction information DB using the determined event-specific variation pattern And a congestion prediction correction unit that outputs a corrected congestion prediction after correcting.

  The congestion prediction method according to the present invention is a congestion prediction method executed by a congestion prediction device that predicts the congestion status of a store as a congestion amount at each time, and the congestion prediction that is a prediction of the congestion status of the store is performed in advance. The stored congestion prediction information DB and at least one event-specific variation pattern that is a variation pattern for the congestion prediction calculated in consideration of one or a plurality of events, and the event-specific variation pattern information stored in advance DB, and an actual congestion amount before a predetermined reference time of the congestion prediction is acquired, an actual variation pattern indicating a variation of the actual congestion amount is calculated, and the calculated actual variation pattern and By comparing the event-specific variation pattern stored in the event-specific variation pattern information DB before the reference time, the event-specific variation pattern that approximates the actual variation pattern is determined, Using a constant is event-specific variation pattern, and outputs the congestion prediction information corrected congestion prediction corrected portion after than the reference time of the DB on the stored congestion forecast, characterized in that.

  A congestion prediction program according to the present invention causes a computer to execute the congestion prediction method described above.

  According to the present invention, highly accurate congestion prediction can be performed.

It is a block diagram of the congestion prediction system which concerns on 1st Embodiment. 3 is a data configuration diagram of a procedure information DB (held by a business system server) according to the first embodiment. FIG. It is a block diagram of the database server and congestion inquiry server which concern on 1st Embodiment. It is a data block diagram of store information DB which concerns on 1st Embodiment. It is a data block diagram of congestion prediction information DB which concerns on 1st Embodiment. It is a data block diagram of congestion pattern DB which concerns on 1st Embodiment. It is a data block diagram of the day congestion pattern DB which concerns on 1st Embodiment. It is a data block diagram of procedure information DB (held by a congestion inquiry server) which concerns on 1st Embodiment. It is a data block diagram of congestion prediction correction information DB which concerns on 1st Embodiment. It is a sequence diagram which shows the operation | movement of the production | generation of the day congestion pattern which concerns on 1st Embodiment. It is a sequence diagram which shows the operation | movement of the determination of the correction | amendment congestion pattern which concerns on 1st Embodiment. It is a sequence diagram which shows the operation | movement of the inquiry of congestion prediction information which concerns on 1st Embodiment. It is a block diagram of the database server and congestion inquiry server which concern on 2nd Embodiment.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
Each figure is only schematically shown so that the present invention can be fully understood. Therefore, the present invention is not limited to the illustrated example. In addition, in each figure, about the same component or the same component, the same code | symbol is attached | subjected and those overlapping description is abbreviate | omitted.

[First embodiment]
≪Configuration of congestion prediction system according to the first embodiment≫
With reference to FIG. 1, the structure of the congestion prediction system 1 which concerns on 1st Embodiment is demonstrated. FIG. 1 is a configuration diagram of a congestion prediction system 1 according to the first embodiment.

  The congestion prediction system 1 includes a business system server 100 installed in a store, a database server 200 installed in a center, and a congestion inquiry server 300 as a congestion prediction device installed in the center. ), A wide area network (WAN), a dedicated line, and the like so as to be communicable. Thereby, the business system server 100, the database server 200, and the congestion inquiry server 300 can transmit data to each other.

The database server 200 is communicably connected to a management terminal 3 owned by the administrator 2 who manages the database server 200 via a LAN, a dedicated line, or the like. Thereby, the management terminal 3 can access the database server 200.
The congestion inquiry server 300 is communicably connected to a customer terminal 6 owned by a customer 5 who inquires about the congestion status of a store via a network 4 such as the Internet or a public switched telephone network (PSTN). The Thereby, the customer terminal 6 can access the congestion inquiry server 300 via the network 4.

  The management terminal 3 is a general-purpose PC (Personal Computer) that is installed in a headquarters or a center and used by an administrator. The management terminal 3 is equipped with client software such as a Web browser, and accesses the congestion prediction management function 320 (see FIG. 3) and the congestion pattern management function 330 (see FIG. 3) included in the congestion inquiry server 300 described later in detail. The congestion prediction information DB 230 (see FIG. 3) is input, and the congestion pattern DB 240 (see FIG. 3) is registered.

The customer terminal 6 is a terminal device such as a general-purpose PC or mobile phone. The customer terminal 6 is equipped with client software such as a Web browser, and accesses the congestion prediction providing function 380 (see FIG. 3) of the congestion inquiry server 300, which will be described later in detail, to make an inquiry about the congestion prediction.
Note that each function described above and below is a functional unit having an entity that is implemented as a program executed or managed by a control unit or the like or a module included in the program. In the description of the present embodiment, the function is simply described for simplification.

<Business system server>
The business system server 100 includes a control unit 110 and a storage unit 150. The control unit 110 is realized by a program execution process by a CPU (Central Processing Unit), a dedicated circuit, or the like. The storage unit 150 includes a storage medium such as a random access memory (RAM), a read only memory (ROM), a hard disk drive (HDD), and a flash memory.

The control unit 110 has a function of receiving a business procedure executed by the store (procedure reception function 120), a function of processing a business procedure executed by the store (processing function 130), and procedure information stored in the procedure information DB 160. Is realized (procedure information providing function 140).
The storage unit 150 stores a procedure information DB 160 that is updated when the procedure receiving function 120 receives a procedure or when the processing function 130 completes processing.

(Procedure Information DB)
FIG. 2 is a data configuration diagram of the procedure information DB 160 according to the first embodiment. The procedure information DB 160 includes a procedure ID 161, a procedure name 162, a received number 163, a completed number 164, and an update time 165. Hereinafter, the procedure ID 161, the procedure name 162, the number of received cases 163, the number of completed cases 164, and the update time 165 may be collectively referred to as procedure information.

In the column of procedure ID 161, identification information for identifying a procedure accepted by the procedure acceptance function 120 is stored. The name of the procedure identified by the procedure ID 161 is stored in the procedure name 162 column. The number of received procedures 163 stores the number of procedures accepted by the procedure accepting function 120. The number of completed cases 164 stores the number of procedures completed by the processing function 130 among the procedures received by the procedure receiving function 120. In the column of update time 165, the last update time of the record identified by the procedure ID 161 is stored in the format “hh (hour): mm (minute)”.
This is the end of the description of the business system server 100.

<Database server>
Next, the configuration of the database server 200 according to the first embodiment will be described with reference to FIG. FIG. 3 is a configuration diagram of the database server 200 and the congestion inquiry server 300 according to the first embodiment.
The database server 200 includes a storage unit 210. The storage unit 210 includes a storage medium such as a random access memory (RAM), a read only memory (ROM), a hard disk drive (HDD), and a flash memory. The storage unit 210 stores a store information DB 220, a congestion prediction information DB 230, a congestion pattern DB 240, a congestion pattern DB 250, a procedure information DB 260, and a congestion prediction correction information DB 270 on the day.

(Store information DB)
Next, the store information DB 220 according to the first embodiment will be described with reference to FIG. FIG. 4 is a data configuration diagram of the store information DB 220 according to the first embodiment. Store information DB220 stores information about stores. The store information DB 220 includes a store ID 221, a store name 222, a business start time 223, and a business end time 224.

  Stored in the store ID 221 column is identification information for identifying the store. The store name 222 stores the store name identified by the store ID 221. In the column of the business start time 223, the time when the store identified by the store ID 221 starts business is stored in the format of “hh (hour): mm (minute)”. Stored in the column of business end time 224 is a time at which the store identified by the store ID 221 ends business.

(Congestion prediction information DB)
Next, the congestion prediction information DB 230 according to the first embodiment will be described with reference to FIG. FIG. 5 is a data configuration diagram of the congestion prediction information DB 230 according to the first embodiment. The congestion prediction information DB 230 stores store congestion predictions.
Here, the “congestion prediction” represents the amount of congestion such as a window or procedure on an arbitrary business day in a store with the passage of time, and constitutes one prediction in a predetermined period. The congestion prediction can be defined by predicting the congestion amount for each time zone from the past congestion situation of the store.

The “congestion amount” is a numerical value of the congestion state and is expressed in a plurality of stages (levels). For example, the “congestion amount” is defined as follows for the procedure congestion status at a store with a store.
In the present embodiment, the congestion amount when the maximum number of procedures that can be processed per unit time is Tmax is set as “Low” / “Medium (Tmid)” / “High” as follows: I will ask.
1. Congestion amount “low”: Tlow <Tmax × 0.5
2. Congestion amount “Tmid”: Tmax × 0.5 ≦ Tmid <Tmax × 0.7
3. Congestion amount “High”: High ≧ Tmax × 0.7
In the following, for convenience, the congestion amount “low” may be represented as “1”, the congestion amount “medium” as “2”, and the congestion amount “high” as “3”.

Using the congestion amount defined above, the congestion forecast xi on a business day of a store is expressed as a set as follows.
xij = {ti1, ti2,..., tij}
Here, i is a variable representing a date, j is a variable representing a time zone, and an element (tij) of the congestion prediction xi is a congestion amount corresponding to the variable i and the variable j. Therefore, the first element (ti1) of the congestion prediction xi represents the amount of congestion at the time of n on a certain day, the second element (ti2) represents the amount of congestion at the time of (n + 1),. The element represents the amount of congestion at (n + j-1) hours.
In the following specific examples, j may be abbreviated as xi.

In the present embodiment, the congestion forecast x1 of the A branch on September 11 is calculated as follows: the congestion amount at 9 o'clock is “low = 1”, the congestion amount at 10 o'clock is “medium = 2”, and the congestion amount at 11 o'clock. Is “high = 3”, the congestion amount at 12:00 is “high = 3”, the congestion amount at 13:00 is “high = 3”, and the congestion amount at 14:00 is “medium = 2”. In that case, j = 6 is omitted, and the congestion prediction x1 on September 11 is as follows.
x1 = {1, 2, 3, 3, 3, 2}

In addition, the congestion forecast x2 of the A branch on September 12 is shown as “medium = 2” at 9 o'clock, “medium = 2” at 10 o'clock, and “medium” at 11 o'clock. = 2 ”, the congestion amount at 12:00 is“ high = 3 ”, the congestion amount at 13:00 is“ low = 1 ”, and the congestion amount at 14:00 is“ low = 1 ”. In that case, the congestion prediction x2 on September 12 is as follows.
x2 = {2, 2, 2, 3, 1, 1}

The congestion prediction information DB 230 includes a store ID 231, a date 232, a time zone 233, and a congestion amount 234.
In the column of store ID 231, identification information for identifying a store is stored. In the date 232 column, a predetermined date on which the store identified by the store ID 231 is scheduled to operate is stored in the format of “MM month DD day”. In the column of the time zone 233, a predetermined time zone of the date 232 that the store identified by the store ID 231 is scheduled to operate is in the format of “hh hour (from hh to (hh + 1) hour before)”. Stored. In the column of the congestion amount 234, a predicted value of the congestion amount in the time zone 233 is stored.

  The record of the congestion prediction information DB 230 corresponds to each element of the congestion prediction xi. For example, in the case of the congestion forecast x1 of the September 11 branch A, the record 2311 represents the first element (t11) of the congestion forecast x1, the record 2312 represents the second element (t12), and the record 2313 represents the third The element 2 represents the element (t13), the record 2314 represents the fourth element (t14), the record 2315 represents the fifth element (t15), and the record 2316 represents the sixth element (t16). Therefore, the congestion prediction x1 is composed of records 2311 to 2316.

(Congestion pattern DB)
Next, the congestion pattern DB 240 according to the first embodiment will be described with reference to FIG. FIG. 6 is a data configuration diagram of the congestion pattern DB 240 according to the first embodiment. The congestion pattern DB 240 as the event-specific variation pattern DB stores a congestion pattern as the event-specific variation pattern.
The “congestion pattern” as the “event-specific variation pattern” is a variation pattern with respect to the congestion prediction xi, and is calculated in consideration of one or a plurality of events. Here, the events considered in calculating the congestion pattern include the weather conditions in the area where the store exists, the characteristics of the area where the store exists, the occurrence of a specific event in the area where the store exists, or the event There is a malfunction of the equipment.

The congestion pattern can be defined by, for example, grasping the variation tendency based on the variation of each store with respect to the past congestion prediction of the store (by counting or the like).
In the present embodiment, “variation” is defined as follows using a difference with respect to the congestion amount of the congestion prediction xi.
1. When the past congestion prediction is “high = 3” while the past actual congestion amount is “low = 1”, the fluctuation is set to “−2” and the past actual congestion amount Is “−1” when “medium = 2”, and “+0” when the past actual congestion amount is “high = 3”.

2. When the past traffic congestion amount is “medium = 2” while the past actual congestion amount is “low = 1”, the fluctuation is set to “−1” and the past actual congestion amount Is “+0” when “medium = 2”, and “+10” when the past actual congestion amount is “high = 3”.
3. When the past congestion forecast is “Low = 1” while the past actual congestion quantity is “Low = 1”, the fluctuation is set to “+0”, and the past actual congestion quantity is When “medium = 2”, the variation is “+1”, and when the actual actual congestion amount is “high = 3”, the variation is “+2”.

Using the variation (difference) defined above, the congestion pattern Km is expressed as a set as follows.
Kmj = {um1, um2, ..., umj}
Here, m is a variable representing the type of the congestion pattern, j is a variable representing the time zone, and the element (umj) of the congestion pattern Kmj is a variation in the congestion amount corresponding to the variable m and the variable j. is there. Therefore, the first element (um1) of the congestion pattern Km represents the fluctuation of the congestion amount in the n hour range specified by the pattern ID m, and the second element (um2) represents the fluctuation of the congestion amount in the (n + 1) hour period. ,..., The j th element represents a change in the amount of congestion at (n + j−1) hours. In the following specific examples, j is sometimes abbreviated as Km.

In the present embodiment, the congestion pattern Ka in which the afternoon procedure is more congested than predicted is represented by “+0” for the fluctuation at 9 o'clock, “−1” for the fluctuation at 10 o'clock, “−1” for the fluctuation at 11 o'clock. It is assumed that the fluctuation at 12:00 is “+1”, the fluctuation at 13:00 is “+2”, and the fluctuation at 14:00 is “+2”. In that case, the congestion pattern Ka is as follows.
Ka = {+ 0, -1, -1, +1, +2, +2}

Also, the congestion pattern Kb, which is more crowded than expected in the morning procedure, is “+1” for 9 o'clock, “+2” for 10 o'clock, “+2” for 11 o'clock, and “12” It is assumed that the fluctuation is “−1”, the fluctuation at 13:00 is “−1”, and the fluctuation at 14:00 is “+0”. In that case, the congestion pattern Kb is as follows.
Kb = {+ 1, +2, +2, -1, -1, +0}

Further, the congestion pattern Kc in which the 10:00 to 13:00 range is congested from the prediction, the change at 9 o'clock is “+0”, the change at 10 o'clock is “+1”, the change at 11:00 is “+2”, 12 It is assumed that the fluctuation of the hour base is “+2”, the fluctuation of the 13:00 hour is “+2”, and the fluctuation of the 14:00 hour is “+0”. In that case, the congestion pattern Kc is as follows.
Kc = {+ 0, + 1, + 2, + 2, + 2, + 0}

The congestion pattern DB 240 includes a pattern ID 241, a time zone 242, and a fluctuation amount 243 of the congestion amount.
In the pattern ID 241 column, identification information for identifying the congestion pattern Km is stored. In the present embodiment, the variation pattern with the pattern ID 241 “Ka” represents a pattern in which the afternoon procedure is more congested than predicted, the variation pattern with the pattern ID 241 “Kb” represents the pattern in which the morning procedure is congested than predicted, The variation pattern whose pattern ID 241 is “Kc” represents a pattern in which the 10: 00-13: 00 range is more congested than predicted. A predetermined time zone is stored in the column of the time zone 242 in the format of “hh hours (from hh to before (hh + 1) hours)”. In this embodiment, values of “−2”, “−1”, “0”, “+1”, and “+2” are stored in the congestion amount variation 243 column. In the following, “variation in congestion” may be simply referred to as “variation”. Further, the “+” notation may be omitted.

  The record of the congestion pattern DB 240 corresponds to each element of the congestion pattern Km. For example, in the case of the congestion pattern Ka, the record 2411 represents the first element (ua1) of the event-specific variation pattern Ka, the record 2412 represents the first element (ua2), and the record 2413 represents the first element (ua3). , Record 2414 represents the first element (ua4), record 2415 represents the first element (ua5), and record 2416 represents the first element (ua6). Therefore, the congestion pattern Ka is composed of records 2411 to 2416.

(Daily congestion pattern DB)
Next, with reference to FIG. 7, the day congestion pattern DB 250 according to the first embodiment will be described. FIG. 7 is a data configuration diagram of the same day congestion pattern DB 250 according to the first embodiment. The same day congestion pattern DB 250 as the actual fluctuation pattern DB stores the same day congestion pattern as the actual fluctuation pattern.
The “daily congestion pattern” as the “actual fluctuation pattern” is a fluctuation pattern with respect to the congestion prediction xi, and is calculated from the actual congestion amount of the store. The configuration of the congestion pattern on the day is the same as the congestion pattern.

For example, the congestion pattern Kt on the day of September 11 of the A branch is “+0” for 9 o'clock fluctuation, “−1” for 10 o’clock, “−1” for 11:00 o’clock, 12:00 When the fluctuation from the base to the 14:00 is set to “null”, the congestion pattern Kt on the day is as follows.
Kt = {+ 0, -1, -1, null, null, null}

Next, the current day congestion pattern DB 250 includes a store ID 251, a date 252, a time zone 253, and a congestion amount variation 254.
In the column of store ID 251, identification information for identifying the store is stored. In the column of the date 252, a predetermined date on which the store identified by the store ID 251 has performed business is stored in the format “MM month DD day”. In the column of the time zone 253, a predetermined time zone of the date 252 on which the store identified by the store ID 251 operates is stored in the format of “hh hour (between hh and (hh + 1) hours)”. Is done. In the congestion amount change 254 column, values of “−2”, “−1”, “0”, “+1”, “+2”, and “null” are stored in this embodiment. “Null” represents an initial value, and indicates that the fluctuation is “uncalculated” because the actual congestion amount is not acquired.

  Further, the record of the current day congestion pattern DB 250 corresponds to each element of the current day congestion pattern Kt. The record 2511 represents the first element (ut1) of the current day congestion pattern Kt, the record 2512 represents the first element (ut2), the record 2513 represents the first element (ut3), and the record 2514 represents the first element (ut4). The record 2515 represents the first element (ut5), and the record 2516 represents the first element (ut6). Therefore, the current day congestion pattern Kt is composed of records 2511 to 2516.

(Procedure Information DB)
Next, the procedure information DB 260 according to the first embodiment will be described with reference to FIG. FIG. 8 is a data configuration diagram of the procedure information DB 260 according to the first embodiment.
The procedure information DB 260 includes a procedure ID 261, a procedure name 262, the number of received cases 263, the number of completed cases 264, and an update time 265. Since the configuration of the procedure information DB 260 is the same as the configuration of the procedure information DB 160, description thereof is omitted.

(Congestion prediction correction information DB)
Next, the congestion prediction correction information DB 270 according to the first embodiment will be described with reference to FIG. FIG. 9 is a data configuration diagram of the congestion prediction correction information DB 270 according to the first embodiment.
The congestion prediction correction information DB 270 includes a store ID 271, a date 272, a time zone 273, and correction information 274.

The store ID field 271 stores identification information for identifying the store. In the date 272 column, a predetermined date on which the store identified by the store ID 271 has worked is stored in the format of “MM month DD day”. In the column of the time zone 273, a predetermined time zone of the date 272 on which the store identified by the store ID 271 has been operated is stored in the format of “hh hours (between hh hours to (hh + 1) hours before)”. Is done. In the correction information 274 column, information for correcting the congestion prediction xi is stored as variation (difference). In this embodiment, the values “−2”, “−1”, “0”, “+1”, and “+2” are stored.
This is the end of the description of the database server 200.

<Congestion inquiry server>
Next, the configuration of the congestion inquiry server 300 according to the first embodiment will be described with reference to FIG. The congestion inquiry server 300 as a congestion prediction device includes a control unit 310.
The control unit 310 is realized by a program execution process by a CPU (Central Processing Unit), a dedicated circuit, or the like. When the congestion inquiry server 300 is realized by program execution processing, a congestion prediction program for realizing the function of the congestion inquiry server 300 is stored in a storage unit (not shown).

  The control unit 310 includes a congestion prediction management function 320, a congestion pattern management function 330, a procedure information collection function 340, a current day congestion pattern management function 350, a corrected congestion pattern determination function 360, a congestion prediction correction function 370, and a congestion. And a prediction providing function 380. Hereinafter, each function will be described in detail.

(Congestion prediction management function)
The congestion prediction management function 320 manages the congestion prediction information DB 230. Specifically, the congestion prediction management function 320 accepts access by the management terminal 3 (see FIG. 1) operated by the administrator 2 (see FIG. 1), and stores the store ID 231, date 232, and time zone input by the administrator 2. Information about the 233 and the congestion amount 234 is reflected in the congestion prediction information DB 230.
Here, it is assumed that the manager 2 updates the congestion prediction information DB 230 in advance based on the past congestion prediction xi that does not take into account individual specific events of the store. For example, the manager 2 classifies the stores into large groups such as urban stores, suburban stores, and regional stores, and performs the congestion prediction xi.

(Congestion pattern management function)
The congestion pattern management function 330 manages the congestion pattern DB 240. Specifically, the congestion pattern management function 330 accepts access from the management terminal 3 (see FIG. 1) operated by the administrator 2 (see FIG. 1), and receives the pattern ID 241 input by the administrator 2, the time zone 242 and Information on the fluctuation 243 in the congestion amount is reflected in the congestion pattern DB 240.

Here, the administrator 2 calculates a congestion pattern Km that is a variation pattern with respect to the congestion prediction xi in consideration of one or a plurality of events, and updates the congestion pattern DB 240 in advance based on the calculated congestion pattern Km. And The events considered for calculating the congestion pattern Km include, for example, the weather conditions in the area where the store exists, the characteristics of the area where the store exists, the occurrence of a specific event in the area where the store exists, or the event set in the store There is a malfunction of the equipment.
For example, the manager 2 calculates the congestion pattern Km by grasping the tendency of the fluctuation based on the fluctuation of each time zone with respect to the past congestion prediction xi of the store (aggregation). Here, the relationship between the event and the congestion pattern Km is not necessarily clear. For this reason, even a variation pattern caused by a plurality of events can be set as the congestion pattern Km.

(Procedure information collection function)
The procedure information collection function 340 collects information such as the procedure of the day from the business system server 100 (see FIG. 1). Specifically, the procedure information collection function 340 receives the contents of the procedure information DB 160 (see FIG. 1) stored in the business system server 100 every predetermined period (for example, one hour), and receives the received contents in the procedure information DB 260. To reflect.

(Daily congestion pattern management function)
The day congestion pattern management function 350 manages the day congestion pattern DB 250. Specifically, the current day congestion pattern management function 350 updates the current day congestion pattern DB 250 every predetermined period (for example, one hour) based on the congestion prediction information DB 230 and the procedure information DB 260 reflected by the procedure information collection function 340. .

For example, the number of waits for reception at 9 o'clock on September 11 at branch A (the number of recipients 263-the number of completions 264) is "9", the number of waits at 10 o'clock is "7", and the wait is 11 o'clock Assume that the number of cases is “13”. Further, it is assumed that the maximum number (Tmax) of processing cases per unit time zone of the window is 20. The congestion pattern management function 350 on the day uses these numerical values to obtain the amount of congestion in each time zone by applying the following definition.
1. Congestion amount “low”: Tlow <Tmax × 0.5
2. Congestion amount “Tmid”: Tmax × 0.5 ≦ Tmid <Tmax × 0.7
3. Congestion amount “High”: High ≧ Tmax × 0.7

On the day of the congestion pattern management function 350, the actual amount of congestion at 9 o'clock in the A branch is “low = 1”, the amount of congestion at 10 o'clock is “low = 1”, and the amount of congestion at 11 o'clock is It is calculated that “medium = 2”.
Next, the current day congestion pattern management function 350 accumulates and manages the current day congestion pattern Kt in the current day congestion pattern DB 250 as information on temporal fluctuation (difference) from the congestion prediction x1. Specifically, the actual day congestion pattern management function 350 determines that the calculated actual 9 o'clock congestion amount “low = 1” and the congestion amount “low = 1” of the record 2311 (see FIG. 5) of the congestion prediction information DB 230. Difference “0” is stored in the fluctuation amount 254 of the record 2511 (see FIG. 7), and the difference between the actual congestion amount “low = 1” in the 10 o'clock range and the congestion amount “medium = 2” of the record 2312 is stored. “−1” is stored in the congestion amount fluctuation 254 of the record 2512, and the difference “−1” between the actual congestion amount “medium = 2” in the 11 o'clock range and the congestion amount “high = 3” of the record 2313 is recorded It is stored in the congestion amount fluctuation 254 of 2513.

As a result, the current day congestion pattern Kt of the branch A at 12:00 on September 11 is expressed as follows.
Kt = {+ 0, -1, -1, null, null, null}

(Correction congestion pattern determination function)
The correction congestion pattern determination function 360 compares the elements of the current day congestion pattern Kt stored in the current day congestion pattern DB 250 and the congestion pattern Km (for example, Ka, Kb, Kc) stored in the congestion pattern DB 240 to match each other. The congestion pattern Km to be calculated is calculated, and the calculated congestion pattern Km is determined as a corrected congestion pattern Kn for correcting the congestion prediction xi. Note that when the current day congestion pattern Kt and the congestion pattern Km are compared, the approximate congestion pattern Km may be determined as the corrected congestion pattern Kn. This degree of approximation can be evaluated, for example, by calculating the inner product of the congestion pattern Kt and the congestion pattern Km on the day. The correction congestion pattern determination function 360 generates congestion prediction correction information using the congestion pattern Km determined as the correction congestion pattern Kn, and updates the congestion prediction correction information DB 270.

  For example, the congestion pattern Kt = {+ 0, −1, −1, null, null, null} on the day, and the congestion pattern Ka = {+ 0, −1, −1, + 1, + 2, + 2}, Kb = {+ 1, + 2 , +2, -1, -1, +0}, Kc = {+ 0, +1, +2, +2, +2, +0}, the first element, the second element, and the third element are compared. Since the first element, the second element, and the third element coincide with the first element, the second element, and the third element of the congestion pattern Ka, the congestion pattern Ka is determined as the corrected congestion pattern Kn. Accordingly, it is predicted that the fourth element, the fifth element, and the sixth element of the day-to-day congestion pattern Kt of the branch A after 12:00 will change as the fourth element, the fifth element, and the sixth element of the congestion pattern Ka. it can. This is because it is considered that the tendency of fluctuations matches (or approximates) due to the occurrence of the same event as the congestion pattern Ka.

  Here, in order to determine the corrected congestion pattern Kn, the elements (the first element, the second element, and the third element in the above) that are compared with the congestion pattern Kt on the day and the congestion pattern Km are called determination target portions. In addition, the elements (the fourth element, the fifth element, and the sixth element in the above description) on which the fluctuation of the congestion pattern Kt on the day is expected based on the corrected congestion pattern Kn may be referred to as a correction target portion. In addition, the boundary time between the determination target part and the correction target part may be referred to as a reference time.

(Congestion prediction correction function)
The congestion prediction correction function 370 uses the corrected congestion pattern Kn (specifically, the correction information 274 (see FIG. 9) stored in the congestion prediction correction information DB 270) to calculate the congestion prediction xi stored in the congestion prediction information DB 230. The corrected congestion prediction xi is output to the customer terminal 6 using the congestion prediction providing function 380 described later. Here, if the corrected congestion amount is less than the minimum congestion amount “low (1)”, the corrected congestion amount is “low (1)” and the maximum congestion amount “high (3)” is set. If it exceeds, set the corrected congestion amount to “High (3)”.
For example, the congestion prediction x1 = {1, 2, 3, 3, 3, 2} is corrected using the corrected congestion pattern Kn = {+ 0, -1, -1, + 1, + 2, + 2}, and the corrected congestion is obtained. Prediction x1 = {Low (1), Low (1), Medium (2), High (3), High (3), High (3)} is output.

(Congestion prediction provision function)
The congestion prediction providing function 380 accepts access by the customer terminal 6 operated by the customer 5, and the congestion prediction x1 corrected by the congestion prediction correction function 370 is transmitted to the customer terminal 6 as data in, for example, HTML (Hypertext Markup Language) format. Send to.
This is the end of the description of the congestion inquiry server 300. Also, the description of the configuration of the congestion prediction system 1 ends.

<< Operation of the congestion prediction system according to the first embodiment >>
Below, the main operation | movement of the congestion prediction system 1 is demonstrated.
<Generation of congestion pattern on the day>
With reference to FIG. 10, the operation | movement of the day congestion pattern production | generation of the congestion prediction system 1 is demonstrated.
The procedure information collection function 340 of the congestion inquiry server 300 is activated every hour using a timer, for example (step S10). Next, the procedure information collection function 340 makes a procedure information acquisition request to the business system server 100 (step S20). Next, the procedure information providing function 140 of the business system server 100 provides procedure information stored in the procedure information DB 160 as a response to the request in step S20 (step S30).

  Next, the procedure information collection function 340 of the congestion inquiry server 300 transmits the procedure information storage request provided in step S30 to the database server 200 (step S40). Next, the database server 200 stores the procedure information in the procedure information DB 260 in response to the request in step S40 (step S50).

Next, the current day congestion pattern management function 350 of the congestion inquiry server 300 calculates the current day congestion pattern Kt (step S60). Next, the current day congestion pattern management function 350 makes a storage request for the current day congestion pattern Kt calculated in step S60 to the database server 200 (step S70). Next, the database server 200 stores the current day congestion pattern Kt in the current day congestion pattern DB 250 in response to the request in step S70 (step S80).
This is the end of the description of the operation for generating the current day congestion pattern.

<Determination of correction congestion pattern>
With reference to FIG. 11, the operation of determining the corrected congestion pattern of the congestion prediction system 1 will be described.
The corrected congestion pattern determination function 360 of the congestion inquiry server 300 is activated every hour using a timer, for example (step S110). Next, the corrected congestion pattern determination function 360 searches the congestion pattern DB 250 on that day (step 120). The database server 200 provides the same day congestion pattern Kt that meets the search condition of step S120 (step S130).

  Next, the corrected congestion pattern determination function 360 searches the congestion pattern DB 240 for a congestion pattern Km that matches the congestion pattern Kt on the day received in step S130 (step S140). Next, the database server 200 provides a congestion pattern Km that meets the search condition of step S140 (step S150).

Next, the correction congestion pattern determination function 360 determines the congestion pattern Km provided in step S150 as the correction congestion pattern Kn (step S160). Next, the corrected congestion pattern determination function 360 generates congestion prediction correction information using the corrected congestion pattern Kn determined in step S160 (step S170). Next, the correction congestion pattern determination function 360 makes a storage request for the congestion prediction correction information generated in step S170 to the database server 200 (step S180). Next, the database server 200 stores the congestion prediction correction information in the congestion prediction correction information DB 270 in response to the request in step S180 (step S190).
This is the end of the description of the operation for determining the correction congestion pattern.

<Inquiry of congestion forecast information>
With reference to FIG. 12, the operation of querying the congestion prediction information of the congestion prediction system 1 will be described.
The congestion prediction correction function 370 of the congestion inquiry server 300 is activated each time in response to a request from the customer 5 (step S210). Next, the congestion prediction correction function 370 searches the congestion prediction information DB 230 (step 220). Next, the database server 200 provides information related to the congestion prediction xi that meets the search condition of Step S220 (Step S230).

Next, the congestion prediction correction function 370 searches the congestion prediction correction information DB 270 (step 240). Next, the database server 200 provides correction information 274 (FIG. 9) that meets the search condition of step S240 (step S250). Next, the congestion prediction correction function 370 corrects the congestion prediction xi received in step S230 using the correction information 274 provided in step S250, and provides the corrected congestion prediction xi as the congestion prediction. Output to the customer terminal 6 using the function 380.
This is the end of the description of the congestion prediction information inquiry operation.

  As described above, the congestion prediction system 1 according to the first embodiment and the congestion inquiry server 300 as the congestion prediction device collect information such as procedures on the day, and automatically calculate the congestion amount based on the collected information. By changing the predicted value, maintenance (correction) of congestion prediction can be performed. Therefore, the congestion prediction system 1 according to the first embodiment and the congestion inquiry server 300 as the congestion prediction device can perform prediction information with higher accuracy.

For example, when a regional event is held in the morning of a specific store, the congestion inquiry server 300 sets the reference time to noon, determines a congestion pattern Km in which the amount of congestion in the morning is reduced, and congestion in the afternoon The predicted value is corrected to increase the amount. On the other hand, when the local event is performed in the afternoon, the congestion inquiry server 300 determines the congestion pattern Km in which the morning congestion amount has increased, and corrects the predicted value so as to reduce the afternoon congestion amount.
By the way, in the case of rain, it is predicted that the amount of congestion in the morning will decrease and the amount of congestion in the afternoon will also decrease. Since the amount of congestion in the morning (difference in the amount of congestion) is different between the case of a regional event held in the morning and the case of rainy weather, the congestion inquiry server 300 can determine an appropriate congestion pattern Km.

[Second Embodiment]
The congestion inquiry server 300 according to the first embodiment automatically corrects the congestion prediction xi using the congestion pattern Km that matches (or approximates) the congestion pattern Kt on the day. On the other hand, the congestion inquiry server 300a according to the second embodiment displays at least one congestion pattern Km that matches (approximates) the congestion pattern Kt on the day together with the probability of approximation (for example, on a terminal in the store). Display) and have the crowd pattern Km be selected. As a result, the congestion inquiry server 300a according to the second embodiment can correct the congestion amount associated with the human feeling.

<Congestion inquiry server according to the second embodiment>
With reference to FIG. 13, a configuration of the congestion inquiry server 300a according to the second embodiment will be described. FIG. 13 is a configuration diagram of the database server and the congestion inquiry server according to the second embodiment.
The difference between the congestion inquiry server 300 according to the first embodiment and the congestion inquiry server 300a according to the second embodiment is that the corrected congestion pattern determination function 360 (see FIG. 3) is changed to the correction candidate congestion pattern determination function 360a. And the congestion prediction correction function 370 (see FIG. 3) is changed to a congestion prediction correction candidate output function 370a. Below, only the function changed from 1st Embodiment among the structures of the congestion inquiry server 300a which concerns on 2nd Embodiment is demonstrated.

(Correction candidate congestion pattern determination function)
The correction candidate congestion pattern determination function 360a calculates the probability that the current day congestion pattern Kt stored in the current day congestion pattern DB 250 and the congestion pattern Km (for example, Ka, Kb, Kc) stored in the congestion pattern DB 240 approximate. The congestion pattern Km having the calculated probability equal to or greater than a predetermined value is determined as a correction candidate congestion pattern Kn that is a candidate for correcting the congestion prediction xi.

(Congestion prediction correction candidate output function)
The congestion prediction correction candidate output function 370a outputs the correction candidate congestion pattern Kn together with the establishment of approximation. Then, the congestion prediction xi is corrected using the selected pattern among the output correction candidate congestion patterns Kn.
Above, description of the congestion inquiry server 300a which concerns on 2nd Embodiment is complete | finished.

  As described above, the congestion inquiry server 300a as the congestion prediction device according to the second embodiment displays at least one congestion pattern Km that approximates the congestion pattern Kt on the day together with the probability of approximation (for example, in a store) And the congestion prediction xi is corrected using the congestion pattern Km selected by the person in charge. As a result, the congestion inquiry server 300a according to the second embodiment can correct the congestion amount associated with the human feeling.

[Modification]
As mentioned above, although embodiment of this invention was described, this invention is not limited to this, It can implement in the range which does not change the meaning. The modification of embodiment is shown below.

(matching)
In the first embodiment and the second embodiment, the corrected congestion pattern Kn and the correction candidate congestion are compared by comparing a plurality of consecutive elements of the current day congestion pattern Kt and a plurality of corresponding consecutive elements of the congestion pattern Km. Although the pattern Kn has been determined, by comparing a comparison between a function represented by a plurality of continuous elements of the congestion pattern Kt on the day and a function represented by a plurality of corresponding continuous elements of the congestion pattern Km, The correction congestion pattern Kn and the correction candidate congestion pattern Kn may be determined.

(Crowded amount)
In the first embodiment and the second embodiment, the congestion amount is represented by three levels (levels) of “low” / “mid” / “high”, but other levels ( For example, five steps or ten steps may be used.

(Congestion inquiry system)
The congestion inquiry system 1 according to the first embodiment and the second embodiment can be applied to any field in which the congestion amount of services such as a store and a procedure in the store is predicted and provided as information. It can also be applied to a mechanism for adjusting the amount of work such as increasing or decreasing the number of windows by referring to the amount of congestion. In that case, the configuration of the congestion inquiry system 1 can be changed as appropriate.

1 Congestion prediction system 100 Business system server 200 Database server 220 Store information DB
230 Congestion prediction information DB
240 Congestion pattern DB (event-specific variation pattern information DB)
250 Congestion pattern DB on the day
260 Procedure Information DB
270 Congestion prediction correction DB
300,300a Congestion inquiry server (congestion prediction device)
320 Congestion prediction management function 330 Congestion pattern management function 340 Procedure information collection function (actual congestion amount acquisition unit)
350 Congestion pattern management function on the day (actual fluctuation pattern calculation unit)
360 Correction congestion pattern determination function (correction variation pattern determination unit)
370 Congestion prediction correction function 380 Congestion prediction provision function 360a Correction candidate congestion pattern determination function (correction candidate fluctuation pattern determination unit)
370a Congestion prediction correction candidate output function xi Congestion prediction Km Congestion pattern (variation pattern according to event)
Kt Congestion pattern on the day (actual fluctuation pattern)
Kn correction congestion pattern (correction variation pattern), correction candidate congestion pattern (correction candidate variation pattern)

Claims (8)

A congestion prediction device that predicts the congestion status of a store as the amount of congestion at each time,
A congestion prediction information DB in which a congestion prediction that is a prediction of the congestion situation of the store is stored in advance;
At least one event-specific variation pattern that is a variation pattern for the congestion prediction calculated in consideration of one or a plurality of events, and a pre-stored event-specific variation pattern information DB,
An actual congestion amount acquisition unit that acquires an actual congestion amount before a predetermined reference time of the congestion prediction;
An actual variation pattern calculation unit that calculates an actual variation pattern indicating variation in the actual congestion amount;
By comparing the calculated actual variation pattern and the event-specific variation pattern stored in the event-specific variation pattern information DB before the reference time, the event-specific variation pattern that approximates the actual variation pattern is determined. A correction variation pattern determination unit;
Using the determined event-specific variation pattern, a congestion prediction correction unit that outputs a corrected congestion prediction that corrects a portion after the reference time of the congestion prediction stored in the congestion prediction information DB;
A congestion prediction device comprising: The event considered in calculating the event-specific variation pattern is:
The weather conditions of the area where the store exists, the characteristics of the area where the store exists, the occurrence of a specific event in the area where the store exists, or the failure of the equipment installed in the store.
The congestion prediction apparatus according to claim 1. The amount of congestion is represented at a plurality of levels according to the congestion situation,
The congestion prediction is represented by a set having the congestion amount as an element,
The event-specific variation pattern and the actual variation pattern are represented by a set having a difference with respect to the congestion amount of the congestion prediction as an element.
The congestion prediction apparatus according to claim 1 or 2, wherein The correction variation pattern determination unit
Comparison between a plurality of continuous elements of the actual variation pattern and a corresponding plurality of corresponding elements of the variation pattern by event, a function represented by a plurality of continuous elements of the actual variation pattern and the variation by event Determine event-specific variation patterns that approximate the actual variation pattern by comparison with a function represented by the corresponding consecutive elements of the pattern,
The congestion prediction apparatus according to claim 3. The actual congestion amount is
Calculated based on the number of people waiting at the store reception desk,
The congestion prediction apparatus according to any one of claims 1 to 4, wherein the congestion prediction apparatus is characterized in that: A congestion prediction device that predicts the congestion status of a store as the amount of congestion at each time,
A congestion prediction information DB in which a congestion prediction that is a prediction of the congestion situation of the store is stored in advance;
At least one event-specific variation pattern that is a variation pattern for the congestion prediction calculated in consideration of one or a plurality of events, and a pre-stored event-specific variation pattern information DB,
An actual congestion amount acquisition unit that acquires an actual congestion amount before a predetermined reference time of the congestion prediction;
An actual variation pattern calculation unit that calculates an actual variation pattern indicating variation in the actual congestion amount;
By comparing the calculated actual variation pattern and the event-specific variation pattern stored in the event-specific variation pattern information DB before the reference time, the actual variation pattern and the event-specific variation pattern are A correction candidate variation pattern determination unit that calculates a probability of approximation, and determines one or more event-specific variation patterns in which the calculated probability is equal to or greater than a predetermined value;
A congestion prediction correction candidate output unit that outputs the determined one or more event-specific variation patterns together with the probability;
A congestion prediction device comprising: A congestion prediction method executed by a congestion prediction device that predicts the congestion status of a store as a congestion amount at each time,
A congestion prediction information DB in which a congestion prediction that is a prediction of the congestion situation of the store is stored in advance;
At least one event-specific variation pattern that is a variation pattern for the congestion prediction calculated in consideration of one or a plurality of events, and a pre-stored event-specific variation pattern information DB,
Obtain the actual congestion amount before the predetermined reference time of the congestion prediction,
Calculating an actual variation pattern indicating variation in the actual congestion amount;
By comparing the calculated actual variation pattern and the event-specific variation pattern stored in the event-specific variation pattern information DB before the reference time, an event-specific variation pattern that approximates the actual variation pattern is determined. ,
Using the determined event-specific variation pattern, outputting a corrected congestion prediction in which a portion after the reference time of the congestion prediction stored in the congestion prediction information DB is corrected;
A congestion prediction method characterized by that.   A congestion prediction program that causes a computer to execute the congestion prediction method according to claim 7.

Images