JP4879796B2 - RF analysis / RF prediction program - Google Patents

RF analysis / RF prediction program Download PDF

Info

Publication number
JP4879796B2
JP4879796B2 JP2007086077A JP2007086077A JP4879796B2 JP 4879796 B2 JP4879796 B2 JP 4879796B2 JP 2007086077 A JP2007086077 A JP 2007086077A JP 2007086077 A JP2007086077 A JP 2007086077A JP 4879796 B2 JP4879796 B2 JP 4879796B2
Authority
JP
Japan
Prior art keywords
movement
information
rf
cell
cell value
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
JP2007086077A
Other languages
Japanese (ja)
Other versions
JP2008243090A (en
JP2008243090A5 (en
Inventor
卓哉 佐久間
Original Assignee
スプリームシステムコンサルティング株式会社
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by スプリームシステムコンサルティング株式会社 filed Critical スプリームシステムコンサルティング株式会社
Priority to JP2007086077A priority Critical patent/JP4879796B2/en
Publication of JP2008243090A publication Critical patent/JP2008243090A/en
Publication of JP2008243090A5 publication Critical patent/JP2008243090A5/ja
Application granted granted Critical
Publication of JP4879796B2 publication Critical patent/JP4879796B2/en
Application status is Active legal-status Critical
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06QDATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/08Logistics, e.g. warehousing, loading, distribution or shipping; Inventory or stock management, e.g. order filling, procurement or balancing against orders
    • G06Q10/087Inventory or stock management, e.g. order filling, procurement, balancing against orders

Description

  The present invention relates to a computer program that supports the sale of merchandise, and more particularly to a computer program that realizes RF analysis and RF prediction.

It is an eternal challenge for merchandise sellers to predict future sales trends from current sales information and to take appropriate sales tactics to maximize customer numbers, sales, or profits. There is RFM analysis as a customer analysis method for segmenting high-quality customers from the current customer purchase behavior and purchase history. This is a method of analyzing customer trends based on factors such as the latest purchase time (R: Recency) of the product purchaser, the cumulative number of purchases (F: Frequency) in a specific period, and the cumulative purchase price (M: Mononety). . By this method, it is considered that more appropriate sales tactics can be taken by segmenting customers.
International Publication No. 02/057973 Pamphlet

  The above-described prior art is a technique that supports performing a predetermined sales strategy according to a segment to which a customer belongs after performing RFM analysis up to now. However, in the prior art, predetermined sales tactics such as direct mail transmission and email distribution are automatically or manually performed, and the result of the sales tactics is the result of subsequent sales, etc. Until feedback is unknown. Since it is impossible to predict the results of sales tactics, there is a problem that the risk of investing in sales tactics such as campaign implementation is high.

  The present invention has been made in view of such circumstances, and an object of the present invention is to provide a technique for predicting a customer's future product purchase behavior by RF analysis and for assisting in examining appropriate sales tactics. .

  The inventor of the present invention focused on the RF analysis based on the latest purchase time (R) of the product purchaser and the cumulative number of purchases (F) in a specific period among the above-mentioned RFM analysis. The inventors have invented the following computer program, thinking that it is possible to predict the future product purchase behavior of customers by this RF analysis.

One embodiment of the present invention relates to an RF analysis / RF prediction program. This computer program is a two-dimensional plane table in which the number of customer purchases is the first axis and the latest purchase time is the second axis, and the computer program supports analysis using an RF table divided into a predetermined number of cells. And a user data reception function for detecting a specified input by a user of a product category and a period of time to be analyzed, an RF table, and a product category, and a cell in the period to be analyzed RF history analysis function for creating value movement information as first cell value movement information, and second cell value for cell value movement information from the time period subject to analysis to the time of creating an RF table An RF prediction function created as movement information, an RF table creation function for creating an RF table at the time of creating an RF table based on the first cell value movement information or the second cell value movement information, and RF The realizing an RF table output function of the screen display, to a computer.
The RF history analysis function includes a product sales history information acquisition function for acquiring product sales history information for a period to be analyzed from a database holding product sales history information, and an individual sales history information based on the product sales history information. From the customer movement record calculation function that calculates the movement record per unit time on the RF table of the customer and the movement record of each customer, the movement record information per unit time of the cell value between individual cells is calculated. A cell value movement record calculation function for creating first cell value movement information.
The RF prediction function includes a cell value movement ratio calculation function for calculating a movement ratio of cell values per unit time between individual cells based on the first cell value movement information, and a movement ratio at each time. A cell value movement characteristic calculation function for creating series information and calculating a movement characteristic for each unit time of cell values between individual cells based on a predetermined evaluation function from each time series information, and a first cell value movement A cell value movement prediction function for calculating movement prediction information per unit time of cell values between individual cells based on the information and movement characteristics, and creating second cell value movement information.

  The “product category” in this specification includes a single product, a plurality of products, and a product group including one or more products and grouped according to their common properties and attributes.

Another aspect of the present invention also relates to an RF analysis / RF prediction program. This computer program is a two-dimensional plane table in which the number of customer purchases is the first axis and the latest purchase time is the second axis, and the computer program supports analysis using an RF table divided into a predetermined number of cells. And a user data reception function for detecting a specified input by a user of a product category and a period of time to be analyzed, an RF table, and a product category, and a cell in the period to be analyzed RF history analysis function for creating value movement information as first cell value movement information, and second cell value for cell value movement information from the time period subject to analysis to the time of creating an RF table An RF prediction function created as movement information, an RF table creation function for creating an RF table at the time of creating an RF table based on the first cell value movement information and the second cell value movement information, The Realizing an RF table output function of surface display, to a computer.
The RF history analysis function includes a product sales history information acquisition function for acquiring product sales history information for a period to be analyzed from a database holding product sales history information, and an individual sales history information based on the product sales history information. From the customer movement record calculation function that calculates the movement record per unit time on the RF table of the customer and the movement record of each customer, the movement record information per unit time of the cell value between individual cells is calculated. A cell value movement record calculation function for creating first cell value movement information.
The RF prediction function includes a cell value movement ratio calculation function for calculating a movement ratio of cell values per unit time between individual cells based on the first cell value movement information, and a movement ratio at each time. A cell for creating the second cell value movement information by creating series information, calculating movement characteristics per unit time of cell values between individual cells based on a predetermined evaluation function from each time series information A value transfer characteristic calculation function.

  According to these configurations, if the RF table creation time is within the analysis target period, the movement information is created based on the merchandise sales history information. If the RF table creation time is after the analysis target period, movement information is created by the prediction function. As a result, it is possible to create an RF table in a range where product sales history information exists, as well as an RF table at the time when product sales history information does not exist. The user can visualize the business situation at the future time from the output RF table, and it becomes easy to examine appropriate sales tactics.

  It should be noted that any combination of the above-described constituent elements and a representation of the present invention converted between a method, an apparatus, a system, etc. are also effective as an aspect of the present invention.

Now that competition among companies is intensifying, it is an urgent task for merchandise sellers to select appropriate sales tactics and increase customer loyalty. There is RFM analysis as a customer analysis method for segmenting customers when selecting an appropriate sales tactic.
The inventor of the present invention, in particular, in the RFM analysis, calls the latest purchase time (hereinafter referred to as “R”) of the product purchaser and the cumulative number of purchases (hereinafter referred to as “F”) in a specific period. We focused on RF analysis that analyzes customer trends using an RF table centered on). As a result of performing a large number of RF analyzes in a large number of product categories, the characteristic that the value of each cell in the RF table (hereinafter referred to as the “cell value”) transitions to another cell can be calculated with a specific evaluation function I found out. The present RF analysis / RF prediction program was invented and disclosed here, with the idea that the customer's future product purchase behavior can be predicted based on the characteristics of this transition.

Here, the RF table is a result of segmenting each element of the merchandise sales history information including the sales performance of various merchandise and information of the customer who is the purchaser into each two-dimensional cell area of RF.
FIG. 1 is an image diagram of an RF table. The horizontal is the number of purchases consisting of 6 columns, and the vertical is the latest purchase time consisting of 13 rows. In the present specification, the column with the number of purchases of one is hereinafter referred to as F1, and the others are similarly described as F2, F3,. In addition, the row where the latest purchase time is 0 to 1 month ago is described as R1, and the others are also described as R2, R3,. And the display of the position of a cell is shown as (F1, R1) as (the number of purchases, the latest purchase time), for example. However, the number of columns / rows of the RF table and the unit time of the latest purchase time in the present specification are arbitrary, and are not particularly limited to specific numbers. In the following description, the unit time is assumed to be January.
Note that each cell value indicated by “...” In FIG. 1 actually displays the total value of the merchandise sales history information in that cell. This is mainly the number of customers, sales and profits, but is not limited to this. For example, text information such as the mode value of customer attribute information belonging to the cell may be used. Hereinafter, the number of customers will be described as an example of the cell value.
Further, although not shown in FIG. 1, the total result of each cell value may be displayed in the RF table or in the vicinity thereof. For example, in the case of FIG. 1, the sales total and / or the total number of customers may be displayed in or near the RF table.

Customers belonging to each cell in the RF table move to other cells over time.
FIG. 2 is a diagram showing a movement image of customer information on the RF table. This figure shows an example in which a customer belonging to a specific cell moves to another cell as the unit time elapses. Here, a certain customer 202 is in the cell (F2, R3). If this customer 202 does not purchase a product for one month, it moves to the cell (F2, R4) immediately below and becomes the customer 204. Further, when the product is not purchased for one month, it moves to the cells (F2, R5) below and becomes the customer 206. If the customer 202 purchases one product, the number of purchases is 3, and the customer 202 moves to the upper right cell (F3, R1) and becomes the customer 208. When the customer 202 purchases two products of the same product category in the RF analysis, the number of purchases is 4 at a stroke, and the customer 202 moves to the cell (F4, R1) and becomes the customer 210. Thus, the cells that can move from a specific cell in the RF table are limited to the cells immediately below that have the same F and R that is one greater than R, or the cells that have R of 1 and F increased from the current value. In other words, the cells to which the cell value can move with the passage of time are limited. The limitation on the movable cell of this cell value will be described later.

  Along with the movement of the customer in the RF table described above, the merchandise sales history information such as the sales amount associated with the customer and the attribute information of the customer are also moved together. Then, when tabulated as a cell value, information associated with the customer belonging to the cell, for example, the purchase amount of each customer belonging to the cell is tabulated. As a result, the total value of the sales amount displayed on the RF also changes with the passage of time. In the actual RF analysis / RF prediction program, only the customer ID may move on the virtual RF table, and the product sales history information associated with the customer ID may be tabulated at the time of tabulation.

(First embodiment)
FIG. 3 is a functional block diagram showing the configuration of the RF analysis / RF prediction apparatus according to the first embodiment of the present invention. In terms of hardware components, these configurations are realized by a CPU of a computer, a memory, a program loaded in the memory, and the like, but here, functional blocks realized by their cooperation are illustrated. Accordingly, those skilled in the art will understand that these functional blocks can be realized in various forms by hardware only, software only, or a combination thereof.

The RF analysis / RF prediction apparatus 100 includes a user interface function 110, an RF table creation function 120, and an RF calculation function 130.
The database 190 holds product sales history information including at least the latest purchase time of the product purchaser and the cumulative number of purchases in a specific period. The database 190 is connected to the RF analysis / RF prediction apparatus 100 by some communication means such as LAN / WAN / Internet. Furthermore, the RF analysis / RF prediction apparatus 100 and the database 190 may exist on the same hardware.

  The user interface function 110 is in charge of processing related to the entire user interface, such as input processing from the user and information display to the user. In the present embodiment, it is assumed that the user interface service 110 provides the user interface service of the RF analysis / RF prediction apparatus 100. As another example, a client application (not shown) may perform input / output with the RF analysis / RF prediction apparatus 100 via a communication network including the Internet. In this case, a communication unit (not shown) receives an instruction from the client application, and transmits information of the RF table created based on the instruction to the client application.

The user interface function 110 includes a user data reception function 112 and an RF table output function 114. The user data reception function 112 receives an input operation by the user. The analysis request information input from the user includes (1) the analysis period start time, (2) the analysis period end time, (3) product category, and (4) RF table creation time, which are the target period of the RF analysis. . The RF table creation time point (4) is a date and time that serves as a reference for calculating the cell value of the RF table displayed on the user's screen as the input result. This date / time may be a date / time between (1) and (2) above or a date / time after (2).
The target period of RF analysis is a discontinuous period, for example, from June to September 2003, from June to September 2004, from June to September 2005, and from June to September 2006. Can also be specified. Thereby, for example, more appropriate RF analysis can be obtained with respect to seasonal products such as fireworks and mosquito coils and products with limited sales period.
The RF table output function 114 displays the RF table created by the RF table creation function 120 described later on the user's screen.

  The RF calculation function 130 creates movement information of cell values in the RF table. The RF calculation function 130 includes an RF history analysis function 140 and an RF prediction function 150. The RF history analysis function 140 acquires merchandise sales history information from the database 190 and creates cell value movement information (hereinafter referred to as “first cell value movement information”) within an analysis period designated by the user. . Based on the first cell value movement information, the RF prediction function 150 performs cell value movement information (hereinafter referred to as “second cell value movement information”) from the analysis period specified by the user to the RF table creation time point. Create). Details of the RF history analysis function 140 and the RF prediction function 150 will be described later.

  FIG. 4 is a diagram showing cell value movement information between cells in the RF table. The contents shown in the figure are finally output by the RF calculation function 130, and are used when the RF table is created by the RF table creation function 120 described later. For example, in December 2005, 100 customers have moved from the (F1, R1) cell to the (F2, R1) cell. If the analysis period end point specified by the user is February 2006 and the RF table creation time is May 2006, the cell history movement information up to February 2006 is stored in the RF history analysis function 140. Created as first cell value movement information. Further, the cell value movement information from March to May 2006 is created by the RF prediction function 150 as the second cell value movement information.

The RF table creation function 120 creates an RF table at the time of RF table creation specified by the user based on the first cell value movement information or the second cell value movement information. For example, when the RF table for May 2006 is created based on the cell value movement information shown in FIG. 4 and the values of the cells (F3, R1) in the RF table are calculated, the RF table is created. The function 120 adds the numerical values of the destination of movement (F3, R1) in the May 2006 column. In this case, the movement amount 18 from (F1, R1), the movement amount 30 from (F2, R1), and the movement amount from other F (1 or 2) cells (not shown) are summed to obtain the cell (F3 , R1). In the RF table, there is no customer who stays in the current cell even after the unit time elapses, and the movement information between cells is shown in FIG. Column, that is, an RF table at an arbitrary time point can be created.
Further, the RF table creation function 120 may calculate the total value of each cell value by, for example, adding each cell value after obtaining each cell value of the RF table. In this case, the RF table output function 114 may display the total value in or near the RF table.

In the present embodiment, the RF table creation function 120 creates an RF table based on the cell value movement information shown in FIG. 4. However, the RF table creation function 120 uses the designated RF table. When the table creation time is included in the range of the analysis target period, it is also possible to create an RF table directly from the product sales history information. For example, the RF table creation function 120 first acquires product sales history information from the analysis start time to the RF table creation time. Next, the number of purchases and the latest purchase time are acquired for each customer from the merchandise sales history information and mapped to the RF table. Here, the number of purchases by the customer is the number of purchases made by the same customer for the same product category in the product sales history information. For the latest purchase time of the customer, the difference between the RF table creation time and the date and time when the customer last took purchase behavior in the purchase history information is applied. As described above, the RF table at the time of creating the designated RF table can be created.
However, as in the present embodiment, the RF table creation function 120 creates an RF table based on the cell value movement information shown in FIG. Regardless of whether it is included in the range, the RF table can be created by the same procedure as described above.

  FIG. 5 is a flowchart showing the flow of processing in the RF analysis / RF prediction apparatus 100 described so far. First, the user data reception function 112 detects designation input of analysis request information by the user (S502). The RF history analysis function 140 acquires product sales history information for the analysis target period designated by the user from the database 190 (S504). The RF history analysis function 140 further analyzes the merchandise sales history information and creates first cell value movement information (S506). Here, the RF calculation function 130 determines whether the RF table creation time point designated by the user is included in the range of the analysis target period (S508). If not included (N in S508), the RF prediction function 150 creates second cell value movement information up to the RF table creation time based on the first cell value movement information (S510). If it is included (Y in S508), the second cell value movement information is not created. The RF table creation function 120 creates an RF table based on the first cell value movement information or the second cell value movement information (S512). Finally, the RF table output function 114 displays the RF table on the user screen (S514).

  FIG. 6 is a detailed functional block diagram of the RF calculation function 130. The figure shows the detailed functions inside the RF history analysis function 140 and the RF prediction function 150. The RF history analysis function 140 includes a merchandise sales history information acquisition function 142, a customer movement record calculation function 144, and a cell value movement record calculation function 146. The RF prediction function 150 includes a cell value movement ratio calculation function 152, a cell value movement characteristic calculation function 154, and a cell value movement prediction function 156. The product sales history information acquisition function 142 acquires product sales history information for the analysis target period designated by the user from the database 190.

The customer movement record calculation function 144 calculates the movement record per unit time on the RF table of each customer based on the merchandise sales history information.
FIG. 7 is a diagram showing a tracking information image of the movement results of customer information on the RF table. This figure shows an image of the movement record on the RF table of a certain customer A calculated by the customer movement record calculation function 144. Customer A shows the travel record for one year since the purchase of the product for the first time in January 2006. When customer A purchases a product for the first time, it appears as a position 702 on the RF. Then, since the product has not been purchased for two months, it moves to position 704. The product is purchased here and moved to position 706. In this way, the position moves to position 708 after one year.
As described above, the customer movement record calculation function 144 tracks the movement history of each customer and finally creates the customer movement record table shown in FIG.

  FIG. 8 is a table for recording the movement record of the customer on the RF table. Specifically, this figure is a customer movement record table in which one customer is one record and the movement history is recorded as (movement source cell position) → (movement destination cell position) every unit time. In the case of customer A, since a product was purchased for the first time in January 2006, there is no source cell in January 2006 and only the destination cell (F1, R1). Since the product was not purchased in February 2006, the source cell is (F1, R1) and the destination cell is (F1, R2). In this way, the customer's movement record is recorded from the start time of the analysis target period to the end time every unit time.

The cell value movement record calculation function 146 calculates the cell value movement record between cells from the customer movement record and creates first cell value movement information. The cell value movement record calculation function 146 counts the number of cells having the same movement source cell and movement destination cell in the same unit time in the customer movement record record table of FIG. For example, in FIG. 8, in addition to the above-described movement history of customer A, the movement history of customer B is described. Customer B is the customer who purchased the product for the first time in December 2005. It moved from cell (F2, R1) to (F2, R2) in May 2006. Here, in the same month, customer A is moving in the same way. In this case, the cell value movement record calculation function 146 counts that there are two movements from the cell (F2, R1) to the cell (F2, R2) as of May 2006.
In this way, the cell value movement record calculation function 146 calculates the movement source information of the movement source cell, the movement destination cell, and the movement, that is, the movement value of the cell value, based on the results for each unit time, and finally shows in FIG. The cell value movement information on the RF table is created for each combination of cells and used as the first cell value movement information.

  So far, the merchandise sales history information acquisition function 142, the customer movement record calculation function 144, and the cell value movement record calculation function 146 included in the RF history analysis function 140 have been described. As described above, the RF history analysis function 140 creates first cell value movement information that is movement information of cell values within an analysis period designated by the user. When the RF table creation time comes after the analysis period, the RF prediction function 150 creates second cell value movement information, which is cell value movement information from the end of the analysis period to the RF table creation time. Hereinafter, each function included in the RF prediction function 150 will be described.

  The cell value movement rate calculation function 152 calculates the cell value movement rate per unit time from the first cell value movement information. For example, in FIG. 4, the number of customers existing in the cell (F1, R1) as of January 2006 is 440. Of these, as of February 2006, 88 moves to the cell (F2, R1) and 319 moves to the cell (F1, R2). That is, the ratio of the destination cells (F2, R1) and (F1, R2) to the source cells (F1, R1) as of February 2006 is about 20% and about 73%, respectively. As described above, the cell value movement rate calculation function 152 converts each cell value of the first cell value movement information shown in FIG. 4 from the number of movements of the customer to the movement rate.

The cell value movement characteristic calculation function 154 calculates the change tendency of the movement ratio per unit time as the movement characteristic from the movement ratio of each cell of the first cell value movement information calculated by the cell value movement ratio calculation function 152. First, the cell value movement characteristic calculation function 154 creates time series information in which a movement ratio between a specific cell and another specific cell is associated with time information.
FIG. 9 is a diagram illustrating a change in the movement ratio of the cell value between specific cells. In the figure, as an example of time-series information, a movement rate 902 between specific cells in each month of 2005 is shown.
Next, the cell value movement characteristic calculation function 154 calculates a movement characteristic used to create the second cell value movement information from the movement ratio 902 based on a predetermined evaluation function. Then, the movement characteristics are calculated for the number of records of the first cell value movement information in FIG. 4, in other words, for the number of pairs of the movement source and the movement destination in the movement of the cell value. For example, mobility characteristics between cell (F1, R1) and cell (F2, R1), mobility characteristics between cell (F1, R1) and cell (F3, R1), and so on. calculate. That is, finally, the cell value movement characteristic calculation function 154 is the movement information of the cell value of the RF table in the format of FIG. 4, and each cell value is set to the movement ratio calculated by the movement characteristic. Output information. For example, if the movement characteristic is a constant having a fixed value between specific cells, the constant is described in the row value between the cells.

The evaluation function used by the cell value movement characteristic calculation function 154 may be a function for calculating an arithmetic average of movement ratios of individual unit times, or a function for calculating a geometric mean, a harmonic average, a mode value, a median value, or the like. A combination of these functions may be used. Further, the movement rate trend may be obtained by other known arithmetic / statistical methods. Unless otherwise specified, when calculating other movement ratios, movement characteristics, and the like in this specification, there is no limit to the value calculation method as in this example.
In addition, this evaluation function may determine the trend of the movement ratio by dividing the analysis target period into specific periods. For example, as a movement characteristic between the cell (F1, R1) and the cell (F2, R1), A% may be the cell value movement rate in January every year, and B% may be the cell value movement rate in February every year. From December to February every year, the movement characteristics may be calculated with C% as the cell value movement ratio and from March to May as D% as the cell value movement ratio.

In addition, this evaluation function may be a function that calculates a convergence value at which the cell value movement rate between cells as the unit time elapses converges. In this case, the cell value movement characteristic calculation function 154 may use the convergence value as a movement ratio constant when calculating the movement characteristic.
The present inventor has found that the cell value movement ratio between cells in the RF table converges to a specific convergence value, and uses this convergence value as a fixed value of the movement ratio. This led to the idea that the movement rate at the future time could be predicted efficiently. In the present specification, “converge” is described as a concept including that the movement ratio during a certain period falls within a predetermined range from a specific value.
Thereby, once the convergence value of the movement ratio is calculated, the second cell value movement information can be created using this convergence value as a constant, and subsequent recalculation of the movement characteristics is not necessary. Therefore, efficient utilization of hardware resources and quick response to user requests are facilitated.

As described above, there is no limitation on the method for calculating the convergence value that becomes the movement characteristic. For example, the convergence value can be obtained by calculating an approximate straight line having a slope of 0 using the least square method.
FIG. 10 is a diagram showing a change in the movement ratio of cell values and movement characteristics between specific cells. This figure shows an example in which the movement characteristic 1002 is calculated as an approximate straight line with a slope of 0 using the least square method based on the movement ratio 902 shown in FIG.

In addition, when the number of products sold increases due to the implementation of a sales promotion campaign, or when the number of products sold decreases due to some trouble, the movement ratio between specific cells may increase or decrease. is there.
FIG. 11 is also a diagram showing fluctuations in the movement ratio of cell values and movement characteristics between specific cells. The figure shows the same movement ratio between cells as in FIGS. 9 and 10 and the monthly movement ratio 1102 in 2006. The number of sales increased due to the effect of the campaign in February and March, and the rate of movement between cells in February, March and April has increased. On the other hand, the number of sales decreased due to the rise in retail prices accompanying the rise in wholesale prices of goods in August, and the rate of movement between cells in August and September has decreased. From this result, for example, when the moving characteristic 1104 is calculated as an approximate straight line with a slope of 0 using the least square method, the value becomes larger than the moving characteristic 1002 in FIG.

In such a case, the cell value movement characteristic calculation function 154 discriminates such a period in which the movement ratio greatly fluctuates as an invalid period, calculates an approximate straight line from the movement ratio in a period other than the invalid period, and moves Characteristics may be obtained. Specifically, the cell value movement characteristic calculation function 154 may acquire period information such as campaigns and troubles from the merchandise sales history information, and may determine these periods as invalid periods or exceed a predetermined threshold. The period of the movement ratio may be determined as an invalid period. In addition, the cell value movement characteristic calculation function 154 may acquire customer information that has been subject to a campaign or trouble from the product sales history information, and exclude those customers from the object for which movement characteristics are to be obtained. In addition, a movement ratio such as a campaign period or a trouble period specified by the user may be set as an invalid period. Furthermore, the invalid period may be directly designated by the user. Information on these invalid periods may be held in an information holding function (not shown) of the RF analysis / RF prediction apparatus 100 and referred to when calculating the movement characteristics. The movement characteristic 1106 is an approximate straight line calculated by removing the movement ratios in February, March, April, August, and September from the movement ratio 1102 in FIG.
Thereby, when calculating the convergence value of the movement ratio, it is possible to eliminate the influence due to the temporary fluctuation of the movement ratio, and it is possible to calculate the convergence value of the movement ratio more appropriate from a long-term viewpoint. .

Furthermore, this evaluation function may calculate the convergence value based on the fact that the cells to which the value of a specific cell on the RF table can move are limited.
As described above, on the RF table, for example, the cells that can move after the unit time of the cell value in the column where F is 1 are limited to the cells immediately below itself or the cells where R is 1 and F is 2 or more. The In other words, a cell to which a specific cell can move is limited to a cell immediately below itself or a cell having the latest purchase date and the latest number of purchases.
Due to this feature of the RF table, the cell value movement characteristic calculation function 154 may apply the evaluation function only between the limited cells to obtain the convergence value. In other words, the amount of calculation for obtaining the convergence value is greatly reduced, and as a result, efficient use of hardware resources and quick response to user requests are facilitated. Specific cell values on the RF table can be moved. The effect of limiting the cells does not appear only in the cell value movement characteristic calculation function 154. The cell value movement information shown in FIG. 4, for example, the cell value movement record calculation function 146, the cell value movement ratio calculation function 152, and the cell value movement prediction function 156 described later can all move the cell value. By limiting the number of cells, the amount of calculation is greatly reduced.

Here, the reduction of the calculation amount will be considered using the RF table of FIG. When there is no restriction on the movement of cells, the total number of cells is 6 × 13, and each cell can be moved from all cells including itself, so the combination of (6 × 13) ^ 2 cells Need to be calculated. Here, “^” represents a power. As a result, the number of cell combinations is 6084.
Next, consider a case where the movable cells are limited. When there is a purchase action, R always moves to a cell with 1; therefore, first, the number of cells that R moves to a cell with 1 is considered. Since (F1, R1) is an initial value, there is no source cell and it is zero. A cell moving to (F2, R1) is a column of cells with F = 1, and a cell moving to (F3, R1) is a cell with F = 1 or 2. As a result, the number of cells that move to a cell with R of 1 is 0, 13, 13 × 2,..., Starting from 1, and is an arithmetic sequence of initial terms 0, terms 6 and tolerance 13. Thus, the sum of the arithmetic sequence is 195. Exceptionally, when F is 6, since there is a movement from each cell where F is 6 to (F6, R1), 13 is added and the total is 208. In principle, a cell with R of 2 or more is moved from a cell having the same F and F and having R smaller by one than its own R. In exceptional cases, a cell having the largest R may move to itself. The combination is 6 × 13, for a total of 78. Therefore, in the case of the RF table of FIG. 1, there are a total of 286 cell combinations, and the number of combinations is significantly smaller than in the case where there is no restriction on cell movement.

The cell value movement prediction function 156 is based on the first cell value movement information and the movement characteristic calculated by the cell value movement characteristic calculation function 154, that is, the cell value movement information at the future time point, that is, the second cell value movement. Create information.
For example, in FIG. 4, it is assumed that the period up to February 2006 is designated as the analysis target period and is given as the first movement information, and the RF table creation time is designated as May 2006. Furthermore, as a result of the cell value movement characteristic calculation function 154, the movement characteristic between cells is given as a fixed value ratio. For example, the movement characteristic between the cell (F2, R1) and the cell (F3, R1) is α%. Suppose that it was given as a move. In this case, the cell value movement prediction function 156 first calculates the cell value of the cell (F2, R1) in February. In this calculation, as in the RF table creation function 120 described above, the values in the February column and the destination cell (F2, R1) are tabulated. Next, a value obtained by multiplying the value of the cell (F2, R1) in February by α% is set as the value of March of the source cell (F2, R1) and the destination cell (F3, R1). This is repeated for each set of source cell and destination cell, and the column for March 2006 in FIG. 4, in other words, the cell value migration information for March is calculated.

Next, the cell value movement prediction function 156 calculates the cell value movement information for April based on the cell value movement information for March, and finally moves the cell value until the RF table creation time. Is calculated. In this example, the movement characteristic is a fixed value, but the movement characteristic is not limited to a fixed value. As described above, the movement characteristics may be different for each unit time.
The value of the cell (F1, R1) may be a value set by the user in advance, or the cell value movement prediction function 156 determines that it is appropriate from the value of the past cell (F1, R1). A value to be set may be set. For example, the arithmetic mean, mode, and median of the cell values of the past cell (F1, R1) may be determined to be appropriate, or the past cell (F1, R1) may be determined using the least square method. A future value of the same cell may be predicted from the value column.
Thus, the cell value movement prediction function 156 calculates the movement source information of the movement source cell, the movement destination cell, and the movement, that is, the cell value based on the prediction for each unit time, and finally the RF value shown in FIG. The cell value movement information on the table is created for each combination of cells and used as second cell value movement information. Based on the second cell value movement information, the RF table creation function 120 creates an RF table at the time of creating the RF table designated by the user.

So far, an example has been described in which the RF analysis / RF prediction apparatus 100 according to the embodiment of the present invention is used to predict the future product purchase behavior of a customer by RF analysis. Here, the flow of processing of each function included in the RF calculation function 130 will be described again using the cell value movement information format of FIG. Here, it is assumed that the end point of the analysis target period is February 2006, and the RF table creation time is May 2006. The cell value movement result calculation function 146 sets the number of customer movements in each cell up to February 2006 as the first cell value information based on the result. The cell value movement ratio calculation function 152 converts each cell value up to February 2006 from a movement number to a movement ratio. The cell value movement characteristic calculation function 154 calculates a movement characteristic from the movement ratio, and sets the movement ratio for each cell from March to May 2006 based on the movement characteristic. The cell value movement prediction function 156 converts the value of each cell from March to May 2006 into the second cell value movement information by converting the movement ratio into the number of movements of the customer. As a result, the RF table creation function 120 can create an RF table at the time of RF table creation based on the information in the May 2006 column. Here, for the sake of simplicity, the description has been made with the image of rewriting the same movement information, but the present invention is not limited to this method.
Hereinafter, as a further specific example, an example in which the RF analysis / RF prediction apparatus 100 assists the user in examining appropriate sales tactics will be described.

In the description so far, in order to calculate a more appropriate movement characteristic in the prediction of the movement ratio in the future, for example, an example of calculating the movement characteristic excluding the influence of the temporary fluctuation of the movement ratio due to the implementation of the campaign. Indicated. Hereinafter, the movement characteristics calculated in this way are referred to as “normal movement characteristics”.
The cell value movement characteristic calculation function 154 may further calculate the movement characteristic of the cell value in the RF table that changes as a result of the campaign as the “campaign movement characteristic”. As described above, the product sales history information may include campaign period information and target customer information, and the cell value movement characteristic calculation function 154 uses this information to move the campaign. The characteristic may be calculated.
Further, the user data reception function 112 may further detect new input or edit by the user of new customer subscription information and campaign information as user simulation information. The cell value movement prediction function 156 may create second cell value movement information based on the user simulation information and the campaign movement characteristic in addition to the first cell value movement information and the normal movement characteristic. The “campaign” in the present specification refers to a sales tactic for sales promotion such as direct mail transmission and email distribution.

As described above, the movement characteristics of the cell values in the RF table can be calculated from the first cell value movement information with a predetermined evaluation function. Further, it has already been described that the convergence value of the movement ratio of the cell value in the first cell value movement information may be used as the movement ratio constant.
Here, factors that cause fluctuations in the movement characteristics include the implementation of a campaign and the participation of new customers. In this example, the second cell value movement information is calculated in consideration of these fluctuation factors.

Usually, when a campaign is executed, there is a high possibility that an existing customer purchases a product again (hereinafter, such a customer is referred to as a “repeater customer”) or purchases a plurality of products at a time. Depending on the customer behavior, the cell value movement information in the RF table differs depending on whether the campaign is performed or not.
The cell value movement characteristic calculation function 154 calculates movement information changed by the campaign from the first cell value movement information, that is, the number of increases and the rate of increase of the cell value movement amount as the campaign movement characteristic. For example, if the cell value movement ratio from the cell (F3, R3) to the cell (F4, R1) is 15% when the campaign is not executed and 25% when the campaign is executed, the normal movement characteristic is set to 15%. The campaign movement characteristic may be 25%. When there are a plurality of campaign types, the cell value movement prediction function 156 may calculate a campaign movement characteristic for each campaign type.

A further specific example regarding the calculation of the campaign movement characteristic will be shown.
FIG. 12 is a diagram showing a change in the movement ratio of cell values between specific cells, movement characteristics, and campaign movement characteristics. The figure shows the same normal movement characteristic as the above-described movement characteristic 1106 calculated by excluding the movement ratio 1102 between specific cells in each month of 2006 as described above and the period affected by the campaign or trouble as an invalid period. 1202 is shown. Here, the cell value movement characteristic calculation function 154 may calculate the campaign movement characteristic 1204 by, for example, evaluating a movement ratio from February to April that is affected by the campaign. Here, an approximate straight line having a slope of 0 with respect to the movement ratio from February to April is shown as the campaign movement characteristic 1204 using the least square method. The campaign period varies depending on the business situation when the campaign is executed. Thus, calculating the campaign movement characteristics as an approximate straight line with a slope of 0 does not affect the campaign regardless of the length of the period. It is useful in that it can be grasped. The campaign movement characteristic shown in this example is hereinafter referred to as “first campaign movement characteristic”.

  In addition, the difference between the movement ratio when the campaign is implemented and the movement ratio in the normal movement characteristics is the largest for a while for a while from the start of the campaign, and gradually decreases, and the normal movement characteristics for a while after the campaign ends. It is thought that it will return to the movement rate of. It is also possible to calculate the campaign movement characteristics more faithfully with respect to such movement ratio fluctuations associated with the campaign. For example, when calculating the campaign movement characteristic from February to April in FIG. 12, the cell value movement characteristic calculation function 154 may use the movement ratio from February to April as the campaign movement characteristic of this campaign as it is. As another example, the cell value movement characteristic calculation function 154 obtains an approximate curve of degree n (n is an integer) using the least square method from the movement ratio from February to April, and the equation of the approximate curve May be the campaign movement characteristic of this campaign. In FIG. 12, a campaign movement characteristic 1206 is shown as an example of this approximate curve. In these examples, as the nature of the campaign, the rate of change changes greatly only immediately after the start of the campaign and immediately returns to the normal rate of travel, or the rate of change is large only immediately after the start of the campaign and immediately before the end of the campaign. This is useful in that it can accurately reflect the nature of the campaign when it has movement characteristics such as changing. The campaign movement characteristic shown in this example is hereinafter referred to as “second campaign movement characteristic”.

In any of the above-described examples, the first campaign movement characteristic and / or the second campaign movement characteristic is held together with information indicating the campaign type in the information holding function (not shown) of the RF analysis / RF prediction apparatus 100. Referenced in the process. Furthermore, the example of proper use according to the character of the campaign movement characteristic is shown. The first campaign movement characteristic is different from the actual movement ratio in the campaign implementation than the second campaign movement characteristic, but has a character that it is not easily affected by the length of the period. The second campaign movement characteristic is that the difference from the actual movement ratio in the campaign implementation is small, but it is easily affected by the length of the period, that is, the difference from the actual movement ratio increases as the period changes. Have. Accordingly, more appropriate campaign movement characteristics may be acquired by taking advantage of the movement characteristics of both campaigns as follows. That is, the information holding function holds the first campaign movement characteristic in association with the campaign type, and the second campaign movement characteristic in association with the campaign type and the campaign period. When acquiring the campaign movement characteristics, refer to the information retention function based on the campaign type and the campaign period, and if there is a second campaign movement characteristic that matches both, acquire it. The first campaign movement characteristic that matches may be acquired.
The “campaign movement characteristic” in this specification includes a first campaign movement characteristic and a second campaign movement characteristic.

  As will be described later, when a campaign limited to a specific cell and / or customer is performed, the cell value movement characteristic calculation function 154 calculates the campaign movement characteristic by limiting the cell and / or customer. May be. In other words, when a campaign limited to a specific cell is executed, the campaign movement characteristics from which the cell is the source of movement are calculated from the customer movement ratio of that cell only for the cell where the campaign is executed. Also good. In addition, when a campaign is executed only for a specific customer, it is limited to the cell to which the customer for whom the campaign was executed belongs, and that cell becomes the source of movement based on the movement ratio of the customers for whom the campaign was executed. Campaign movement characteristics may be calculated. Thereby, the calculation amount in the cell value movement characteristic calculation function 154 is reduced, and the response time to the user's request is improved.

  The user data reception function 112 receives new customer subscription information by the user as user simulation information. Specifically, the user can set the number of new customer subscriptions in the cells (F1, R1) of the RF table. The user data reception function 112 receives user campaign information as user simulation information. Specifically, the campaign information includes a plurality of campaign types, each campaign period, and each campaign cost. As the campaign period, the start time and end time of the campaign may be designated. Further, the user may newly input a campaign movement characteristic, or the user may edit the campaign movement characteristic calculated by the cell value movement characteristic calculation function 154.

Further, as described in the description of the user interface function 110 above, the client application may input the user simulation information to the user data reception function 112. In other words, the user data receiving function 112 may receive new customer subscription information and / or campaign information from other systems / devices and the like without human intervention. The client application can realize communication with the RF analysis / RF prediction apparatus 100 according to the present embodiment by using, for example, an API (Application Program Interface) provided by the RF analysis / RF prediction program.
As described above, the client application automates the input of the user simulation information, thereby realizing cooperation between systems and between apparatuses. As a result, the burden on the user is reduced, there is no delay due to human work, and accurate information input without errors can be realized.

  The cell value movement prediction function 156 sets the number of new customer subscriptions specified by the user simulation information in the cell (F1, R1). In addition, for the campaign period specified by the user simulation information, instead of the normal movement characteristic, the movement prediction information is calculated based on the campaign movement characteristic related to the type of campaign specified by the user simulation information, and the second cell Create value transfer information.

  According to this example, first, the user's intention can be reflected in the cell value of the cell (F1, R1), and the effect of the new customer subscription can be visualized. For example, by increasing the number of new customers by 100 from the monthly average so far, it is possible to grasp the number of customers in a specific cell after one year and the increase in sales associated with the customer. Next, the effect of the campaign that the user is planning to implement can be grasped before the campaign is implemented. The effects of the campaign include, for example, an increase in the number of repeater customers and an increase in sales. Furthermore, since the cost of the campaign can be specified in the user simulation information, the profit-based effect by the campaign, that is, the difference between the increase in sales and the campaign cost can be visualized. Thereby, the user can examine the campaign to be executed and the cost-effectiveness of each campaign based on the simulation before actually executing the campaign. In other words, the risk can be reduced prior to the implementation of the campaign.

  In this example, only the campaign movement characteristic is described as a movement characteristic different from the normal movement characteristic, but various modifications such as “increase in retail price due to increase in wholesale price” and “decrease in production due to employee strike” For other business events, the movement characteristic (hereinafter referred to as “event movement characteristic”) may be calculated in the same manner as the campaign. Although the campaign movement characteristic and the event movement characteristic are different in the reason that the movement ratio varies, both are common in that they have a period and a movement ratio, and can be realized by the same procedure as described above. As a result, various events that occur in the user's business, such as the movement ratio that has changed with the event that the retail price has increased due to an increase in the wholesale price and the movement ratio that has changed with the event that the production volume has decreased due to employee strikes. RF analysis / prediction considering the above becomes possible.

  Further, the above-described campaign information may include identification information of one or more cells that are targets of executing each campaign and / or identification information of a plurality of customers that are targets of executing each campaign. That is, a campaign can be set by designating all users belonging to a specific cell at a specific time and / or individual users belonging to any cell at a specific time. These identification information can be set for each campaign. The plurality of customers may be customers belonging to different cells in the RF table.

  When the campaign information includes identification information of a cell to be campaigned, the cell value movement prediction function 156 applies the campaign movement characteristic only to the cell corresponding to the identification information, and the other cells Normal movement characteristics apply. Such a designation is made when a campaign is executed for all customers belonging to a specific cell.

  When the campaign information includes identification information of one or more customers to be campaigned, the cell value movement prediction function 156 applies the campaign movement characteristics only to the customers corresponding to the identification information, The normal travel characteristics apply to customers. For example, assume that there are 100 customers in the cell (F3, R3) at the start of the campaign. When the identification information of 60 of them is designated, the campaign movement characteristic is applied to the 60 persons, and the normal movement characteristic is applied to the remaining 40 persons.

  Note that the same campaign can be applied to customers in a plurality of cells. That is, 60 people of the cell (F3, R3) may be designated as a target of a certain campaign, and 40 people of (F3, R4) may be designated as a target of the same campaign. The campaign movement characteristic of each cell is applied to the designated customer. In this case, the campaign cost is proportionally distributed according to the number of people specified in each cell. In the above example, if the total cost is 100,000 yen, the campaign cost in the cell (F3, R3) is 100,000 yen x 60/100, which is 60,000 yen, and the campaign cost in the cell (F3, R4) Is 40,000 yen in the same calculation. For example, it is assumed that campaign movement characteristics are applied by carrying out a campaign from the cell (F3, R3), and 40 people have purchased goods and moved to the cell (F4, R1). When the campaign is not implemented, the normal movement characteristics are applied and five people purchase. The increase will be 35 people, and if the product unit price is 10,000 yen, the sales growth from the campaign will be 350,000 yen. When converted to profit base, profit growth from the 60,000 yen campaign for 60 people in the cell (F3, R3) is 350,000 yen-60,000 yen, which is 290,000 yen.

  In this way, it is possible to grasp the effect of a campaign limited to a specific cell and / or customer. Since the campaign cost increases when a large-scale campaign is executed, the campaign is often executed only for some customers in an actual business. In this example, campaign information corresponding to the implementation of such a campaign can be set. That is, the user can set a campaign according to the actual business and can grasp the effect. As a result, user convenience can be further enhanced.

  In addition, the RF table creation function 120 may include recommendations for new customer subscriptions to maximize or exceed a particular cell value or cell value in a particular region of the RF table, and Recommendation information regarding the campaign may be created. Here, the recommended information is new customer subscription information and / or campaign information to be set by the user for realizing the above-described conditions such as maximizing a specific one or more cell values.

  In the above example, the second cell value movement information is created based on the subscription information and campaign information of the new customer set by the user, and the RF table creation function 120 is based on the second cell value movement information. An RF table was created. In this example, conversely, the RF table creation function 120 recommends new customer subscriptions and campaigns to maximize or exceed a certain cell value or cell value in a specific area of the RF table. Create information. The predetermined threshold value may be set by the user in advance, or a threshold value calculation unit (not shown) of the RF analysis / RF prediction apparatus 100 may generate an RF table within the analysis target period based on the first cell value movement information. The maximum value of the cell value may be determined as a threshold value.

When maximizing the value of a specific cell, the cell value movement prediction function 156 temporarily sets a plurality of campaigns based on the first cell value movement information, the normal movement characteristic, and the campaign movement characteristic, or newly Temporarily setting the number of subscribers of the customer, the movement prediction information is calculated based thereon, and a plurality of second cell value movement information is created. The RF table creation function 120 creates a plurality of RF tables based on the plurality of second cell value movement information. Next, the RF table creation function 120 specifies the second cell value movement information related to the RF table that satisfies the condition for maximization of the specific cell value, and the campaign information and / or the second cell value movement information Or, the subscription information of a new customer is used as recommended information.
Even when the sum of the cell values in the specific area of the RF table is maximized, only the difference in which the sum of the cell values in the specific area of the RF table is targeted in the determination by the RF table creation function 120 will be described later. Is the same. Also, when the threshold value is equal to or greater than the predetermined threshold, only the difference in specifying the second cell value movement information related to the RF table that satisfies the condition that is equal to or greater than the predetermined threshold in the determination by the RF table creation function 120 is the same. It is.

  Finally, the RF table output function 114 may display an RF table that satisfies a condition that maximizes a specific cell value on the user screen. Further, the recommended information related to the RF table, that is, the new customer subscription information and the campaign information may be displayed so as to call the user's attention. For example, the recommended information may be highlighted by changing a character font, or the recommended information may be displayed in a region different from the RF table on the user's screen.

  According to this example, since the RF analysis / RF prediction apparatus 100 automatically provides recommended information on sales tactics, it is possible to reduce the load on the user who examines appropriate sales tactics. Furthermore, it is possible to provide sales tactics that the user has not thought of. For example, in a particular cell that the user did not consider as the target of the campaign, the campaign response is actually very large, that is, the campaign movement characteristic is very large. You can also recommend having multiple sales tactics. Furthermore, in combination with the previous example, once the recommended information on sales tactics by the RF analysis / RF prediction apparatus 100 is obtained, the campaign execution information is edited based on the user's rule of thumb, and the effect prediction is repeated. May be. The user can further consider effective sales tactics by the recommended information by the RF analysis / RF prediction apparatus 100 and the correction based on the user's experience.

(Second Embodiment)
In the first embodiment, when the RF table creation time is after the analysis target period, the cell value movement prediction function 156 uses the cell value movement information shown in FIG. 4 as the second cell value movement information. The RF table creation function 120 creates an RF table based on the second cell value movement information. In the second embodiment, the cell value movement characteristic calculation function 154 calculates the movement characteristic of each cell value between individual cells, creates second cell value movement information, and creates an RF table creation function. 120 creates an RF table at the time of creating the RF table based on the first cell value movement information and the second cell value movement information.
The second cell value movement information of the present embodiment is an output of the cell value movement characteristic calculation function 154 of the first embodiment. That is, the movement information of the cell values of the RF table in the format of FIG. 4, and each cell value from the end of the analysis target period to the RF table creation time has a movement ratio calculated based on the movement characteristics. It is set.
The RF table creation function 120 of the present embodiment first creates an RF table at the end of the analysis target period based on the first cell value movement information. Next, based on each cell value in the RF table and the movement ratio of the second cell value movement information, an RF table after one unit time has elapsed is created. Next, an RF table after the elapse of 2 unit time is created based on each cell value and movement rate of the RF table after the elapse of 1 unit time. By repeating this, the RF table at the time of RF table creation specified by the user is created.

FIG. 13 is a flowchart showing the flow of processing for creating an RF table after the elapse of unit time from the current RF table and movement characteristics. This figure shows an example of an algorithm for creating an RF table after one unit time has elapsed from an RF table in the format of FIG. In this flowchart, a specific cell value in the current RF table is referred to as “cell value (F i , R j )” and a specific cell value in the RF table after one unit time has elapsed (hereinafter referred to as “future cell value”). ) As a “future cell value (F i , R j )”, and the movement rate between the cell (F i , R j ) and the cell (F m , R n ) in the unit time is expressed as “movement rate (F i , R n )”. j) (F m, are described as R n) ". Note that i, j, m, and n are integers indicating the positions of cells in the RF table.

First, the F-axis loop is entered (S1302), and then the R-axis loop is entered (S1304). Next, a variable x that holds the sum of the number of purchasers of a specific cell is initialized with 0 (S1306). Then, a customer enters a purchase loop for the number of cells in which a customer of a specific cell purchases a product and moves (S1308). Then, it is determined whether the cell value F is 6 (S1310).
When F is not 6 (N in S1310), the movement ratio between a specific cell (F i , R j ) and a cell in which F is greater than i and R is 1 (hereinafter referred to as “purchased moving cell”). Is substituted into the variable y (S1312). Next, the product of the cell value (F i , R j ) and the variable y, that is, the movement amount is calculated and substituted into the variable z (S1314). Next, the variable z is added to the future cell value of the moving cell at the time of purchase (S1316).
When F is 6 (Y in S1310), the movement ratio between a specific cell (F i , R j ) and a cell having the same F and R of 1 (that is, cell (F 6 , R 1 )) is obtained and variable Substitute for y (S1318). Next, the product of the cell value (F i , R j ) and the variable y, that is, the movement amount is calculated and substituted into the variable z (S1320). Next, the variable z is added to the future cell value (F 6 , R 1 ) (S1322).

Next, the variable z is added to the variable x (S1324). This is repeated for the number of mobile cells at the time of purchase (S1326). Up to this point, the total amount of movement from the cell (F i , R j ) to each mobile cell at the time of purchase and the amount of movement to the mobile cell at the time of purchase has been calculated. Here, it is determined whether or not the cell value R is 13 (S1328).
If R is not 13 (N of S1328), the cell (F i, R j) and the current value of, by obtaining the difference between the variable x is the moving amount of the sum of the purchase when the mobile cell, the cell (F i , R j ), the amount of movement to a cell immediately below, that is, a cell with a large R is calculated. Then, the calculated difference is substituted into the future cell value (F i , R j + 1 ) (S1330).
When R is 13 (Y in S1328), the difference between the current value of the cell (F i , R j ) and the variable x, which is the sum of the amount of movement to the purchase movement cell, is obtained to the own cell. Is calculated. The calculated difference is added to the future cell value (F i , R 13 ), that is, its own future cell value (S1332). This is repeated for the number of rows on the R axis (S1334), and further for the number of columns on the F axis (S1336).
As a result, an RF table after one unit time has elapsed can be created. In order to create an RF table at the time of RF table creation, the process shown here is repeated every unit time according to the RF creation time.

As is clear from the above description, the difference between the first embodiment and the second embodiment is that the RF table is created at the time when the RF table is created when the RF table creation time is after the analysis target period. Is the method. That is, in the first embodiment, the cell value movement prediction function 156 creates the number of movements of the cell value up to the RF table creation time, and the RF table creation function 120 calculates the number of movements of the cell value at the time of RF table creation. An RF table can be created directly. On the other hand, in the second embodiment, the cell value movement characteristic calculation function 154 creates the movement ratio of the cell value up to the RF table creation time, and the RF table creation function 120 uses the movement ratio as a basis. The RF table after the elapse of the unit time is sequentially created, and finally the RF table at the time of creating the RF table can be created.
Therefore, it is obvious to those skilled in the art that features that are not directly related to RF table creation are the same in any embodiment, and can be realized in the same manner as in the first embodiment also in this embodiment. A similar effect can be obtained.
Hereinafter, the difference between this embodiment and the first embodiment will be described.

  Similarly to the first embodiment, the cell value movement characteristic calculation function 154 of the present embodiment may further calculate the movement characteristic of the cell value of the RF table that changes due to the implementation of the campaign as the campaign movement characteristic. Here, the cell value movement characteristic calculation function 154 of the present embodiment further includes a unit of cell values between individual cells based on the first cell value movement information, the normal movement characteristic, and the campaign movement characteristic. Second cell value movement information may be created by calculating a movement ratio for each time.

  As described above, the cell value movement characteristic calculation function 154 creates a cell value movement rate up to the time of RF table creation. Here, for the campaign period specified in the user simulation information, the cell value movement characteristic calculation function 154 replaces the normal movement characteristic with the campaign movement characteristic related to the campaign type specified in the user simulation information. Set as the value move rate. Then, the RF table creation function 120 creates an RF table at the time of RF table creation based on the first cell value movement information and the second cell value movement information as described above. In the cell (F1, R1), the RF table creation function 120 sets the number of new customer subscriptions specified by the user simulation information.

  Further, as in the first embodiment, the RF table creation function 120 further maximizes or sets a specific cell value or a total value of cell values in a specific region of the RF table to a predetermined threshold value or more. , Recommendations regarding new customer subscriptions and / or recommendations regarding campaigns may be created. At this time, the cell value movement characteristic calculation function 154 according to the present embodiment temporarily sets a plurality of campaigns based on the first cell value movement information, the normal movement characteristic, and the campaign movement characteristic, and based on the settings, sets a plurality of campaigns. Is calculated, and a plurality of pieces of second cell value movement information are created. Thereafter, the RF table creation function 120 creates an RF table by the method of this embodiment described above. The processing after the condition determination in the RF table creation function 120 is the same as in the first embodiment.

  In the description of any of the embodiments, for the sake of simplification of description, no mention is made regarding cancellation, which is an exceptional event. When cancellation occurs, the customer moves on the RF table to any cell where F is smaller than the current cell. However, even when cancellation occurs, when the first cell value movement information is calculated, it is only necessary to calculate the movement result of the customer on the assumption that there is no purchase behavior related to the cancellation. Further, in calculating the second cell value movement information, since the first cell value movement information has already been processed, it is not necessary to consider cancellation. That is, it can be said that the destination cell of the customer on the RF table is substantially limited as described above. Therefore, it is obvious to those skilled in the art that even merchandise sales history information including cancellation information is an object of analysis of the present invention, and that the present invention can provide RF analysis / RF prediction that also considers cancellation performance.

The present invention has been described based on the embodiments. This embodiment is an exemplification, and it will be understood by those skilled in the art that various modifications can be made to combinations of the respective constituent elements and processing processes, and such modifications are also within the scope of the present invention. is there.
It should be understood by those skilled in the art that the functions to be fulfilled by the constituent elements described in the claims are realized by the individual function blocks shown in the present embodiment or their linkage.

It is a figure which shows the image of RF table. It is a figure which shows the movement image of the customer information on RF table. It is a functional block diagram of the RF analysis / RF prediction apparatus according to the embodiment. It is a figure which shows the movement information of the cell value between the cells of RF table. It is a flowchart which shows the flow of a process in RF analysis and RF prediction apparatus. It is a detailed functional block diagram of RF calculation function. It is a figure which shows the tracking information image of the movement track record of the customer information on RF table. It is a table | surface which records the movement performance on a customer's RF table. It is a figure which shows the fluctuation | variation of the movement ratio of the cell value between specific cells. It is a figure which shows the fluctuation | variation of the movement ratio of a cell value between specific cells, and a movement characteristic. It is a figure which shows the fluctuation | variation of the movement ratio of a cell value between specific cells, and a movement characteristic. It is a figure which shows the fluctuation | variation of the movement ratio of a cell value between specific cells, a movement characteristic, and a campaign movement characteristic. It is a flowchart which shows the flow of the process which produces RF table after unit time progress from the present RF table and a movement characteristic.

Explanation of symbols

  RF analysis / RF prediction apparatus 100, user interface function 110, user data reception function 112, RF table output function 114, RF table creation function 120, RF calculation function 130, RF history analysis function 140, merchandise sales history information acquisition function 142 , Customer movement performance calculation function 144, cell value movement performance calculation function 146, RF prediction function 150, cell value movement ratio calculation function 152, cell value movement characteristic calculation function 154, cell value movement prediction function 156, 190 database, 202 customer, 204 customer, 206 customer, 208 customer, 210 customer, 702 position, 704 position, 706 position, 708 position, 902 movement ratio, 1002 movement characteristic, 1102 movement ratio, 1104 movement characteristic, 1106 movement characteristic, 1202 normal movement characteristic, 1204 Cat Pane transfer characteristics, 1206 campaign movement characteristics.

Claims (12)

  1. A computer program that supports analysis by an RF table divided into a predetermined number of cells, a two-dimensional plane table with the number of customer purchases as the first axis and the latest purchase time as the second axis,
    A user data reception function for detecting, as analysis request information, a period to be analyzed, a point in time when an RF table is created, and a product category user designation input;
    RF history analysis function for creating movement information of the cell value within the period to be analyzed as first cell value movement information;
    An RF prediction function for creating, as second cell value movement information, movement information of the cell value from a period after the period to be analyzed to a point in time when the RF table is created;
    An RF table creation function for creating an RF table at the time of creating the RF table based on the first cell value movement information or the second cell value movement information;
    An RF table output function for displaying the RF table on a screen;
    Is realized on a computer,
    The RF history analysis function
    A product sales history information acquisition function for acquiring product sales history information for a period to be analyzed from a database holding product sales history information;
    A customer movement record calculation function for calculating a movement record per unit time on the RF table of each customer based on the product sales history information;
    A cell value movement record calculation function for calculating the movement value information for each unit time of the cell value between the individual cells from the movement record of the individual customer, and creating the first cell value movement information;
    Including
    The RF prediction function
    Based on the first cell value movement information, a cell value movement ratio calculation function for calculating a movement ratio per unit time of cell values between individual cells;
    Each time series information of the movement ratio is created, a convergence value at which the movement ratio of the value between the cells with unit time elapses is calculated from each time series information, and the convergence value is a value between the cells. Cell value movement characteristic calculation function that specifies as a movement characteristic that indicates a constant of the movement rate per unit time of,
    Based on the first cell value movement information and the movement characteristic, a cell that calculates movement prediction information for each unit time of a cell value between individual cells and creates the second cell value movement information Value movement prediction function,
    Only including,
    The cell value movement characteristic calculation function further sets the movement characteristic as a normal movement characteristic. Separately, the cell value movement characteristic calculation function calculates a convergence value of a movement ratio per unit time of cell values between individual cells during the campaign period. As
    The user data reception function further detects, as user simulation information, new input or edit by a user or client application of new customer subscription information and / or campaign information,
    The cell value movement prediction function is provided in the user simulation information in order to reflect the effect when a new customer specified in the user simulation information is realized and / or when a campaign is executed in the RF table. Accordingly , a computer program characterized in that the movement prediction information is calculated by selectively using the normal movement characteristic or the campaign movement characteristic .
  2.   The computer program according to claim 1, wherein the time point at which the RF table is created is after a period to be analyzed.
  3. The computer program according to claim 1 , wherein the cell value movement characteristic calculation function calculates the convergence value based on a limitation of a cell to which a cell value on the RF table can move.
  4. The cell to which the value of the cell on the RF table can move is a cell having the latest purchase time one larger and the same number of purchases, or a cell having the latest purchase time most recently and the number of purchases increased. The computer program according to claim 3 .
  5. The cell value movement characteristic calculation function further determines that a part of the time series information is an invalid period, and calculates the movement characteristic based on a movement ratio other than the invalid period. The computer program according to any one of 1 to 4 .
  6. The computer program according to claim 5 , wherein the invalid period includes a period specified by a user and / or a campaign period.
  7. The campaign information, one or more the type of campaign, and each campaign period, a computer program according to any of the cost of each campaign claim 1, characterized in that it comprises a 6.
  8. The campaign information further includes identification information of one or more cells to be executed for each campaign and / or identification information of a plurality of customers to be executed for each campaign. 8. The computer program according to 7 .
  9. The RF table creation function further includes recommendations for new customer subscriptions for maximizing or exceeding a predetermined threshold value for a specific cell value or a cell value in a specific region of the RF table, and / or computer program according to any of claims 1 to 8, characterized in that to create the recommendation information related to the campaign.
  10. A computer program that supports analysis by an RF table divided into a predetermined number of cells, a two-dimensional plane table with the number of customer purchases as the first axis and the latest purchase time as the second axis,
    A user data reception function for detecting, as analysis request information, a period to be analyzed, a point in time when an RF table is created, and a product category user designation input;
    RF history analysis function for creating movement information of the cell value within the period to be analyzed as first cell value movement information;
    An RF prediction function for creating, as second cell value movement information, movement information of the cell value from a period after the period to be analyzed to a point in time when the RF table is created;
    An RF table creation function for creating an RF table at the time of creating the RF table based on the first cell value movement information and the second cell value movement information;
    An RF table output function for displaying the RF table on a screen;
    Is realized on a computer,
    The RF history analysis function
    A product sales history information acquisition function for acquiring product sales history information for a period to be analyzed from a database holding product sales history information;
    A customer movement record calculation function for calculating a movement record per unit time on the RF table of each customer based on the product sales history information;
    A cell value movement record calculation function for calculating the movement value information for each unit time of the cell value between the individual cells from the movement record of the individual customer, and creating the first cell value movement information;
    Including
    The RF prediction function
    Based on the first cell value movement information, a cell value movement ratio calculation function for calculating a movement ratio per unit time of cell values between individual cells;
    Each time series information of the movement ratio is created, a convergence value at which the movement ratio of the value between the cells with unit time elapses is calculated from each time series information, and the convergence value is a value between the cells. A cell value movement characteristic calculation function that specifies the movement characteristic indicating a constant of the movement rate per unit time and creates the second cell value movement information;
    Only including,
    The cell value movement characteristic calculation function further sets the movement characteristic as a normal movement characteristic. Separately, the cell value movement characteristic calculation function calculates a convergence value of a movement ratio per unit time of cell values between individual cells during the campaign period. As
    The user data reception function further detects, as user simulation information, new input or edit by a user or client application of new customer subscription information and / or campaign information,
    The cell value movement characteristic calculation function is configured to reflect the effect when a new customer specified in the user simulation information is joined and / or when a campaign is executed in the RF table. According to the above, as the movement characteristic for each unit time, the information in which the normal movement characteristic or the campaign movement characteristic is selectively set is created as the second cell value movement information,
    The RF table creation function is configured to create an RF table after a unit time elapses based on the second cell value movement information in the creation of the RF table from after the period to be analyzed to when the RF table is created. Are sequentially created, and an RF table at the time of creating the RF table is created .
  11. This is a two-dimensional plane table with the customer's purchase count as the first axis and the latest purchase time as the second axis. There,
    A user data reception unit for detecting a period to be analyzed, an RF table creation time point, and a designation input by a user of a product category as analysis request information;
    An RF history analysis unit that creates movement information of the cell value within the period to be analyzed as first cell value movement information;
    An RF prediction unit that creates movement information of the cell value as second cell value movement information from the time period to be analyzed to the time of creating the RF table;
    An RF table creation unit that creates an RF table at the time of creating the RF table based on the first cell value movement information or the second cell value movement information;
    An RF table output unit for displaying the RF table on a screen;
    With
    The RF history analysis unit
    A product sales history information acquisition unit for acquiring product sales history information for a period to be analyzed from a database holding product sales history information;
    A customer movement record calculation unit for calculating a movement record per unit time on the RF table of each customer based on the product sales history information;
    From the movement record of the individual customer, cell value movement record calculation unit for calculating the movement record information for each unit time of the cell value between the individual cells, and creating the first cell value movement information,
    Including
    The RF prediction unit
    Based on the first cell value movement information, a cell value movement ratio calculation unit that calculates a movement ratio of cell values per unit time between individual cells;
    Each time series information of the movement ratio is created, a convergence value at which the movement ratio of the value between the cells with unit time elapses is calculated from each time series information, and the convergence value is a value between the cells. A cell value movement characteristic calculation unit that specifies a movement characteristic indicating a constant of a movement ratio per unit time ,
    Based on the first cell value movement information and the movement characteristic, a cell that calculates movement prediction information for each unit time of a cell value between individual cells and creates the second cell value movement information A value movement prediction unit;
    Only including,
    The cell value movement characteristic calculation unit further sets the movement characteristic as a normal movement characteristic. Separately, the cell value movement characteristic calculation unit calculates a convergence value of a movement ratio per unit time of cell values between individual cells during the campaign period. As
    The user data reception unit further detects, as user simulation information, new input or edit of subscription information and / or campaign information of a new customer by a user or a client application,
    The cell value movement prediction unit may add to the user simulation information in order to reflect the effect when the new customer specified in the user simulation information is joined and / or when the campaign is executed on the RF table. Therefore , the RF analysis / RF prediction apparatus characterized in that the movement prediction information is calculated by selectively using the normal movement characteristic or the campaign movement characteristic .
  12. This is a two-dimensional plane table with the customer's purchase count as the first axis and the latest purchase time as the second axis. There,
    A user data reception unit for detecting a period to be analyzed, an RF table creation time point, and a designation input by a user of a product category as analysis request information;
    An RF history analysis unit that creates movement information of the cell value within the period to be analyzed as first cell value movement information;
    An RF prediction unit that creates movement information of the cell value as second cell value movement information from the time period to be analyzed to the time of creating the RF table;
    An RF table creation unit for creating an RF table at the time of creating the RF table based on the first cell value movement information and the second cell value movement information;
    An RF table output unit for displaying the RF table on a screen;
    With
    The RF history analysis unit
    A product sales history information acquisition unit for acquiring product sales history information for a period to be analyzed from a database holding product sales history information;
    A customer movement record calculation unit for calculating a movement record per unit time on the RF table of each customer based on the product sales history information;
    From the movement record of the individual customer, cell value movement record calculation unit for calculating the movement record information for each unit time of the cell value between the individual cells, and creating the first cell value movement information,
    Including
    The RF prediction unit
    Based on the first cell value movement information, a cell value movement ratio calculation unit that calculates a movement ratio of cell values per unit time between individual cells;
    Each time series information of the movement ratio is created, a convergence value at which the movement ratio of the value between the cells with unit time elapses is calculated from each time series information, and the convergence value is a value between the cells. A cell value movement characteristic calculation unit that specifies the movement characteristic indicating a constant of the movement rate per unit time and creates the second cell value movement information;
    Only including,
    The cell value movement characteristic calculation unit further sets the movement characteristic as a normal movement characteristic. Separately, the cell value movement characteristic calculation unit calculates a convergence value of a movement ratio per unit time of cell values between individual cells during the campaign period. As
    The user data reception unit further detects, as user simulation information, new input or edit of subscription information and / or campaign information of a new customer by a user or a client application,
    The cell value movement characteristic calculation unit is configured to reflect the effect obtained when a new customer specified in the user simulation information is joined and / or when a campaign is executed in the RF table. According to the above, as the movement characteristic for each unit time, the information in which the normal movement characteristic or the campaign movement characteristic is selectively set is created as the second cell value movement information,
    The RF table creation unit is configured to create an RF table after a unit time has elapsed, based on the second cell value movement information, in creating an RF table from a period after the analysis target period to a time when the RF table is created. Is created sequentially, and an RF table at the time of creating the RF table is created .
JP2007086077A 2007-03-28 2007-03-28 RF analysis / RF prediction program Active JP4879796B2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2007086077A JP4879796B2 (en) 2007-03-28 2007-03-28 RF analysis / RF prediction program

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2007086077A JP4879796B2 (en) 2007-03-28 2007-03-28 RF analysis / RF prediction program
PCT/JP2008/000689 WO2008117532A1 (en) 2007-03-28 2008-03-24 Rf analysis/rf prediction program, and rf analysis/rf prediction device

Publications (3)

Publication Number Publication Date
JP2008243090A JP2008243090A (en) 2008-10-09
JP2008243090A5 JP2008243090A5 (en) 2010-05-13
JP4879796B2 true JP4879796B2 (en) 2012-02-22

Family

ID=39788274

Family Applications (1)

Application Number Title Priority Date Filing Date
JP2007086077A Active JP4879796B2 (en) 2007-03-28 2007-03-28 RF analysis / RF prediction program

Country Status (2)

Country Link
JP (1) JP4879796B2 (en)
WO (1) WO2008117532A1 (en)

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2002259672A (en) * 2001-02-26 2002-09-13 Nippon Telegr & Teleph Corp <Ntt> Hybrid type demand and/or profit estimate method
JP2002358402A (en) * 2001-05-31 2002-12-13 Dentsu Tec Inc Method for sales predicting based upon customer value by three index axes
JP2003022359A (en) * 2001-07-06 2003-01-24 Hitachi Ltd Method and device for analyzing customer lifetime value
JP2004054780A (en) * 2002-07-23 2004-02-19 Dainippon Printing Co Ltd Customer management method by ir transition analysis
JP2004185598A (en) * 2002-09-19 2004-07-02 Ricoh Co Ltd One-to-one business support system and program and recording medium for realizing functions of the system

Also Published As

Publication number Publication date
JP2008243090A (en) 2008-10-09
WO2008117532A1 (en) 2008-10-02

Similar Documents

Publication Publication Date Title
Pauwels et al. The long-term effects of price promotions on category incidence, brand choice, and purchase quantity
Hill et al. Network-based marketing: Identifying likely adopters via consumer networks
Zhao et al. Modeling consumer learning from online product reviews
US7848946B2 (en) Sales history decomposition
Ratliff et al. A multi-flight recapture heuristic for estimating unconstrained demand from airline bookings
US20130124257A1 (en) Engagement scoring
US20150040038A1 (en) Configurable computation modules
DE10257199A1 (en) Optimized pricing plan generation method for business items, involves determining mathematical model comprising set of initial constraints, and representing pricing plan for group of item
US20140108640A1 (en) Anomaly Detection in Network-Site Metrics Using Predictive Modeling
US20120158485A1 (en) Integrated and comprehensive advertising campaign management and optimization
US20110282712A1 (en) Survey reporting
US20130060595A1 (en) Inventory management and budgeting system
WO2005024689A1 (en) Nsumer’s purchase behavior analysis method and device
US9147198B2 (en) Systems and methods for providing an interface for data driven media placement
JP2012524340A (en) Travel price optimization (TPO)
US20130166379A1 (en) Social Targeting
Raman et al. Optimal resource allocation with time-varying marketing effectiveness, margins and costs
US9396444B2 (en) Predictive analytics with forecasting model selection
Hwang et al. Joint demand and capacity management in a restaurant system
CN104954410A (en) Message pushing method, device thereof and server
US20110276392A1 (en) Performing Geography-Based Advertising Experiments
JP4465417B2 (en) Customer segment estimation device
US9183562B2 (en) Method and system for determining touchpoint attribution
US20140289071A1 (en) Automatic product groupings for merchandising
US20080262903A1 (en) System and method to determine the prices and order quantities that maximize a retailer&#39;s total profit

Legal Events

Date Code Title Description
A621 Written request for application examination

Free format text: JAPANESE INTERMEDIATE CODE: A621

Effective date: 20100325

A521 Written amendment

Free format text: JAPANESE INTERMEDIATE CODE: A523

Effective date: 20100325

TRDD Decision of grant or rejection written
A01 Written decision to grant a patent or to grant a registration (utility model)

Free format text: JAPANESE INTERMEDIATE CODE: A01

Effective date: 20111122

A01 Written decision to grant a patent or to grant a registration (utility model)

Free format text: JAPANESE INTERMEDIATE CODE: A01

A61 First payment of annual fees (during grant procedure)

Free format text: JAPANESE INTERMEDIATE CODE: A61

Effective date: 20111130

R150 Certificate of patent or registration of utility model

Ref document number: 4879796

Country of ref document: JP

Free format text: JAPANESE INTERMEDIATE CODE: R150

Free format text: JAPANESE INTERMEDIATE CODE: R150

FPAY Renewal fee payment (event date is renewal date of database)

Free format text: PAYMENT UNTIL: 20141209

Year of fee payment: 3

S531 Written request for registration of change of domicile

Free format text: JAPANESE INTERMEDIATE CODE: R313531

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250

R350 Written notification of registration of transfer

Free format text: JAPANESE INTERMEDIATE CODE: R350

S533 Written request for registration of change of name

Free format text: JAPANESE INTERMEDIATE CODE: R313533

S531 Written request for registration of change of domicile

Free format text: JAPANESE INTERMEDIATE CODE: R313531

R350 Written notification of registration of transfer

Free format text: JAPANESE INTERMEDIATE CODE: R350

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250