JP2016006621A - Purchase prediction analysis system and program therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a purchase prediction analysis system for predicting a commodity purchase ratio in a region where the commodity purchase ratio cannot be properly acquired, and a program therefor.SOLUTION: Commodity purchase ratio calculation means 11 reads a number of customers who have purchased a specific commodity and a number of customers, for every cluster, from an ID POS information storage part 22 based on cluster information of a cluster information storage part 21, and calculates a commodity purchase ratio for every cluster. Commodity purchase ratio prediction means 12 determines adoption/non-adoption of the commodity purchase ratio for every cluster, sets the commodity purchase ratio for every region as an objective variable about the region corresponding to the adopted cluster, sets a factor score as an explanatory variable for creating a computation expression of a prediction model of multiple regression analysis, and predicts the commodity purchase ratio about the region corresponding to the cluster of adoption or non-adoption, using the factor score of the region and the computation expression. Analysis result display means 13 reflects the commodity purchase ratio to map information of a map information storage part 24 and displays the commodity purchase ratio, in a purchase prediction analysis system.

Description

  The present invention relates to a system for analyzing purchase forecasts, and more particularly to a purchase forecast analysis system capable of predicting a product purchase rate including a region where a product purchase rate cannot be obtained and a program thereof.

[Conventional technology]
As related technical literature for analyzing conventional purchase forecasts, Japanese Patent Laid-Open No. 10-207958 “Sales Analysis Device and Program Recording Medium” (Tsubasa System Co., Ltd.) [Patent Literature 1], Japanese Patent Laid-Open No. 2002-259672 “Hybrid Type” Demand and / or revenue forecast method "(Nippon Telegraph and Telephone Corporation) [Patent Document 2], Japanese Patent Application Laid-Open No. 2002-358398" Consumption Trend Analysis System and Consumption Trend Analysis Method "(Business Brain Showa Ota Showa) [Patent Document 3], JP 2013-239160 A, “Information providing system, information providing method, and information providing program” (Shiseido Co., Ltd.) [Patent Document 4].

  In Patent Document 1, in the sales analysis device, an area where the store is located is divided into address blocks, and address block groups whose travel time to the store is shorter than the travel time to the other store of the residence are identified, It describes that a map that can identify the amount of sales in the address block is displayed.

  In Patent Literature 2, in a method for predicting demand and profit, a framework of a prediction scenario is determined, a selection behavior model is constructed according to the framework, a selection behavior is predicted and aggregated using the selection behavior model, and the demand of the product is calculated. The prediction is described.

  In Patent Document 3, in a consumption trend analysis system, order data for each group in which an arbitrary region is set as one group is generated based on a customer master database and an order detail database, and analysis of consumption trends can be executed based on the group order data. It is described.

  According to Patent Document 4, in an information providing system, based on information obtained from a store terminal and a Web server, it is integrated into a plurality of hierarchies or a plurality of groups of databases, and the information obtained from them is combined for analysis, and provided by a customer. It is described that a product is determined.

  Although it is not an invention related to purchase prediction, an invention that performs principal component analysis on regional characteristics and performs cluster analysis on the regional characteristics is disclosed in Japanese Patent No. 5284548, “Profiling System Using Regional Characteristics” (GIKEN Shoji International Co., Ltd.) [ Patent Document 5].

JP-A-10-207958 JP 2002-259672-A JP 2002-358398 A JP 2013-239160 A Japanese Patent No. 5284548

  However, in the conventional system for forecasting purchases described above, the number of customers managed by stores in a specific region (ID POS [number of customers with customer ID with POS (Point of Sales)]) and the number of customers who purchased a specific product (product purchase) Customers can calculate the product purchase rate (product purchase customer number / ID POS customer number), but the ID POS customer number and product purchase customer number may not be obtained in all regions, and even if obtained, it is not suitable as a sample Has a problem that it cannot be used as it is as a product purchase rate.

  Further, even if the product purchase rate is obtained by the conventional system for performing purchase prediction, etc., there is a problem that the product purchase rate suitable for the characteristics of each region cannot be predicted.

  The present invention has been made in view of the above circumstances, and an object thereof is to provide a purchase prediction analysis system capable of predicting a product purchase rate for each region including a region where a product purchase rate cannot be obtained properly and a program thereof. To do.

(Purchase forecast analysis system that forecasts product purchase rate)
The present invention for solving the problems of the conventional example described above is a purchase prediction analysis system having an information processing device for predicting a product purchase rate in a region, and stores cluster information obtained by cluster analysis and cluster classification for the region. A cluster information storage unit, a customer information storage unit that stores customer information for stores in a specific area within the analysis target area, and a product purchase rate storage unit that stores a product purchase rate in the specific area. As a function realization means realized by executing the processing program in the control unit of the apparatus, the number of customers from the customer information storage unit based on the cluster information stored in the cluster information storage unit or the total purchase price of the customer and the specific product The product purchase rate stored in the product purchase rate storage unit is calculated by calculating the product purchase rate for each cluster. For the area corresponding to the cluster in which the product purchase rate is adopted by deciding whether or not to adopt the product purchase rate for each cluster using a specific threshold for the calculated product purchase rate and the calculated product purchase rate Using the product purchase rate of each product as an objective variable and the factor score used for cluster analysis as an explanatory variable, a calculation formula for a prediction model for multiple regression analysis is generated to handle clusters in which the product purchase rate is adopted or not adopted And a product purchase rate prediction means for calculating and predicting a product purchase rate using the factor score and calculation formula of the region and storing the product purchase rate in a product purchase rate storage unit.

(Purchasing forecast analysis system server)
The present invention relates to a purchase prediction analysis system having an information processing server that predicts a product purchase rate in a region, a cluster information storage unit that stores cluster information that is cluster-analyzed and classified into regions, and a region to be analyzed A product purchase rate storage unit that stores a product purchase rate in a specific area of the information processing unit, and as a function realization means realized by executing a processing program in the control unit of the information processing device, the analysis target from the client via the network Receive customer information for stores in a specific area within the area, and from the customer information based on the cluster information stored in the cluster information storage unit, the number of customers for each cluster or the total purchase price of the customer and the specific product Read the number of product purchasers or the total purchase price of the product purchase customer, calculate the product purchase rate for each cluster, and calculate the product purchase rate Product purchase rate calculation means to be stored in the memory and a cluster in which the product purchase rate is adopted by determining whether to use or not adopt the product purchase rate for each cluster using a specific threshold for the calculated product purchase rate Using the product purchase rate for each region as an objective variable and the factor score used for cluster analysis as an explanatory variable, a formula for the prediction model of multiple regression analysis is generated and the product purchase rate is adopted or not It has a product purchase rate prediction means for calculating and predicting the product purchase rate using the factor score and calculation formula of the region for the region corresponding to the adopted cluster and storing it in the product purchase rate storage unit. Features.

(Details of adoption / non-adoption of product purchase rate)
According to the present invention, in the purchase prediction analysis system, the product purchase rate calculation means may calculate the number of customers or the total purchase price of the customer and the number of product purchasers of the specific product or the number of the product purchase customer from the customer information. Read the total purchase amount and divide the number of product purchasers by the number of customers, or divide the total purchase amount by the total purchase amount to calculate the product purchase rate for each cluster, and the product purchase rate prediction means The product purchase effective value for each cluster is obtained by multiplying the number of customers or the total purchase price and the product purchase rate. If the product purchase effective value is equal to or greater than a specific threshold, the product purchase rate is adopted and the product purchase effective When the value is less than a specific threshold, the product purchase rate is not adopted.

(Purchase forecast analysis is also performed in areas other than the analysis target area)
According to the present invention, in the purchase prediction analysis system, the product purchase rate storage unit stores a product purchase rate in a region other than the analysis target region, and the product purchase rate prediction unit is stored in the cluster information storage unit. Cluster information is read for regions outside the analyzed area, and the product purchase rate is calculated using the factor score of the area and the formula for the prediction model of the generated multiple regression analysis. The product purchase rate storage unit stores the product purchase rate in the region.

(Contents of analysis result display means / map not used)
In the purchase prediction analysis system, the present invention reads the predicted product purchase rate of the region from the product purchase rate storage unit as a function realization means realized by executing the processing program in the control unit of the information processing device, It is characterized by having an analysis result display means for displaying the product purchase rate or the characteristics of the factors used in the cluster analysis for the products having different values.

(Contents of analysis result display means / map use)
The present invention provides the purchase forecast analysis system as a function realization unit that is provided with a map information storage unit that stores map information related to a region, and that is realized by executing a processing program in the control unit of the information processing apparatus. Analysis result that reads the predicted product purchase rate of the region from the part, reads the map information of the corresponding region from the map information storage unit, and changes the display form of the corresponding region in the map information according to the product purchase rate for each region It has a display means.

(Change the display color or shade of the analysis result display means)
The present invention is characterized in that, in the purchase prediction analysis system, the analysis result display means changes the display color of the corresponding region in the map information or the shade of the display color according to the product purchase rate.

(The product purchase rate of the analysis result display means displays a specific value or more)
The present invention is characterized in that, in the purchase forecast analysis system, the analysis result display means displays a corresponding area in the map information in a special form for an area where the product purchase rate is a specific value or more.

(Displays the store position using the analysis result display means)
The present invention is characterized in that, in the purchase forecast analysis system, the analysis result display means displays the location information of the store in the corresponding area in the map information.

(Purchase forecast analysis program that predicts product purchase rate)
The present invention relates to a purchase prediction analysis program that operates on an information processing device that predicts a product purchase rate in a region, and the information processing device is subjected to cluster analysis for a region stored in a cluster information storage unit, and the cluster is classified into clusters. Based on the information, from the customer information storage unit, for each cluster, the number of customers for the store in the specific area in the analysis target area, or the total purchase price of the customer and the number of product purchasers of the specific product, or the total purchase of the product purchase customer Product purchase rate calculation means that reads the amount of money, calculates the product purchase rate for each cluster, and stores it in the product purchase rate storage unit, and the product purchase rate for each cluster using a specific threshold for the calculated product purchase rate For regions that correspond to clusters for which the product purchase rate is adopted, the product purchase rate for each region is used as an objective variable for cluster analysis. Using the obtained factor score as an explanatory variable, a calculation formula for a prediction model for multiple regression analysis is generated, and for the region corresponding to the cluster where the product purchase rate is adopted or not adopted, the factor score and calculation formula for that region are used. The product purchase rate is calculated and predicted, and functions as a product purchase rate prediction unit that stores the product purchase rate in a product purchase rate storage unit.

(Purchasing forecast analysis system server program)
The present invention relates to a purchase prediction analysis program that operates on an information processing server that predicts a product purchase rate in a region. The information processing server is connected to a store in a specific region within a region to be analyzed from a client via a network. The number of customers for a store in a specific area in the analysis target area for each cluster from the customer information based on the cluster information clustered for the area received information and stored in the cluster information storage unit or Product purchase rate calculation means for reading the total purchase price of the customer and the number of product purchasers of the specific product or the total purchase price of the product purchase customer, calculating the product purchase rate for each cluster, and storing it in the product purchase rate storage unit And adopting or not adopting the product purchase rate for each cluster using a specific threshold for the calculated product purchase rate. For the region corresponding to the cluster where the purchase rate is adopted, the product purchase rate for each region is the objective variable, the factor score used in the cluster analysis is the explanatory variable, and the formula for the prediction model of the multiple regression analysis is generated, The product purchase rate that is predicted by calculating the product purchase rate using the factor score and calculation formula of the region for the region corresponding to the cluster for which the product purchase rate is adopted or not adopted, and storing it in the product purchase rate storage unit It functions as a predicting means.

(Program for determining whether or not to use product purchase rates)
According to the present invention, in the purchase forecast analysis program, the product purchase rate calculation means calculates the number of customers or the total purchase price of the customer and the number of product purchasers of the specific product or the number of the product purchase customer from the customer information. Read the total purchase amount and divide the number of product purchasers by the number of customers, or divide the total purchase amount by the total purchase amount to calculate the product purchase rate for each cluster, and the product purchase rate prediction means The product purchase effective value for each cluster is obtained by multiplying the number of customers or the total purchase price and the product purchase rate. If the product purchase effective value is equal to or greater than a specific threshold, the product purchase rate is adopted and the product purchase effective When the value is less than a specific threshold, the product purchase rate is not adopted.

(Program that conducts purchase forecast analysis in areas other than the analysis target area)
According to the present invention, in the purchase prediction analysis program, the product purchase rate prediction means reads the cluster information for a region other than the analysis target region stored in the cluster information storage unit, and generates the factor score of the region and the generated multiple regression analysis The product purchase rate is calculated and predicted using the calculation model of the prediction model, and is stored in the product purchase rate storage unit as the product purchase rate in a region other than the analysis target region.

(Program that makes the analysis result display function work / No map used)
According to the present invention, in the purchase prediction analysis program, the information processing apparatus or the information processing server reads the predicted product purchase rate in the region from the product purchase rate storage unit, and analyzes the product purchase rate or cluster analysis for products with different conditions. It functions as an analysis result display means for displaying the characteristics of the factors used in the above in a form for comparison.

(Use of program / map to function analysis result display means)
In the above purchase prediction analysis program, the information processing apparatus or information processing server reads the predicted product purchase rate of the region from the product purchase rate storage unit, and reads the map information of the corresponding region from the map information storage unit , And functioning as analysis result display means for changing the display mode of the corresponding region in the map information in accordance with the product purchase rate for each region.

(Change the display color or shade of the analysis result display means)
The present invention is characterized in that, in the purchase prediction analysis program, the analysis result display means changes the display color of the corresponding region in the map information or the shade of the display color according to the product purchase rate.

(The product purchase rate of the analysis result display means displays a specific value or more)
The present invention is characterized in that, in the purchase prediction analysis program, the analysis result display means displays a corresponding area in the map information in a special form for an area where the product purchase rate is a specific value or more.

(Displays the store position using the analysis result display means)
The present invention is characterized in that, in the purchase prediction analysis program, the analysis result display means displays the location information of the store in the corresponding area in the map information.

(Purchase forecast analysis system that forecasts product purchase rate)
According to the present invention, the merchandise purchase rate calculation means can purchase the number of customers or the total purchase price of the customer and the merchandise purchase of the specific merchandise from the customer information storage section based on the cluster information stored in the cluster information storage section. The number of customers or the total purchase price of the product purchase customer is read, the product purchase rate for each cluster is calculated and stored in the product purchase rate storage unit, and the product purchase rate predicting means specifies a specific product purchase rate for the calculated product purchase rate. Use threshold value to decide whether to use or not use the product purchase rate for each cluster, and use the product purchase rate for each region as a target variable for cluster analysis, and use it for cluster analysis. Using the obtained factor score as an explanatory variable, a calculation formula for a prediction model for multiple regression analysis is generated, and for the region corresponding to the cluster where the product purchase rate is adopted or not adopted, the factor score for that region is calculated. The product purchase rate is calculated and calculated using the formula, and the purchase forecast analysis system stores the product purchase rate in the product purchase rate storage unit. There are predictable effects.

It is a schematic block diagram of this system. It is the schematic which shows the structure of the process means and memory | storage part in this system. It is a figure which shows the table | surface of cluster code provision for every administrative boundary. It is a figure which shows the table | surface of a goods purchase rate. It is a figure which shows the table | surface of the cluster code and goods purchase rate for every administrative boundary. It is a figure which shows the table | surface which predicted the goods purchase rate for every administrative boundary. It is a figure which shows the example of a display of an analysis result. It is the schematic of the purchase prediction analysis system of network utilization. It is a figure which shows a goods purchase rate comparison table. It is a spider chart of the factor characteristic comparison of goods purchase. It is a figure which shows the comparison table of a high-order and a low-order area.

Embodiments of the present invention will be described with reference to the drawings.
[Outline of the embodiment]
The purchase prediction analysis system according to the embodiment of the present invention performs cluster analysis on regions with explanatory variables of factors 1 to 9, calculates a product purchase rate (real value) for each cluster, and adopts the calculated product purchase rate / Non-recruitment is determined, the adopted product purchase rate is set in the cluster area where the product purchase rate is adopted, the product purchase rate in the relevant region is set as the objective variable, and the factors 1 to 9 of the relevant region are set. The multiple score analysis formula (prediction model) is generated using the factor score of the model as an explanatory variable, and the product purchase rate (theoretical value) is calculated from the local explanatory variable for all regions using the generated formula. The product purchase rate can be properly predicted by applying cluster analysis and multiple regression formulas in all regions, including regions where product purchase rates cannot be obtained properly. .

[This system: Fig. 1]
A purchase forecast analysis system (present system) according to an embodiment of the present invention will be described with reference to FIG. FIG. 1 is a schematic configuration diagram of the system.
As shown in FIG. 1, the system includes a control unit 1, a storage unit 2, and an interface unit 3 in the information processing apparatus. The interface unit 3 includes a display device 4, an input device 5, The network is connected.
The information processing apparatus in this system uses a general computer, and realizes a function by operating a processing program.

[Parts of this system]
Each part of this system will be specifically described.
[Control unit 1, storage unit 2: FIG. 2]
A specific configuration of the control unit 1 and the storage unit 2 in this system will be described with reference to FIG. FIG. 2 is a schematic diagram showing the configuration of the processing means and the storage unit in this system.

[Storage unit 2]
In addition to the processing program, the storage unit 2 includes a cluster information storage unit 21, an ID POS information storage unit 22, a product purchase rate storage unit 23, and a map information storage unit 24.

[Cluster information storage unit 21]
The cluster information storage unit 21 stores cluster information obtained by cluster analysis and cluster classification for regions divided into administrative boundaries or meshes. The cluster information is obtained, for example, by the technique shown in Patent Document 5, and a cluster code is assigned to the cluster classified characteristics. Note that the control unit 1 may perform the same processing as in Patent Document 5 to generate cluster information and store it in the storage unit 2.

[ID POS information storage unit 22]
The ID POS information storage unit 22 stores customer information in a specific store, and the number of customers (ID POS customers) in the administrative world or mesh and the number of customers who have purchased a specific product (number of products purchased) ) Is remembered. The specific store may have a plurality of stores.
Further, the ID POS information storage unit 22 stores the total purchase price (ID POS total purchase price) of the customer in the administrative circle or mesh and the total purchase price (total purchase price) of the product purchaser who has purchased a specific product. The product purchase rate may be obtained. The total purchase amount is the total amount purchased by the entire number of ID POS customers, and the total purchase amount is the total amount of purchase of the specific product by the entire number of product purchase customers.

[Product purchase rate storage 23]
The product purchase rate storage unit 23 stores the number of ID POS customers and the number of product purchase customers or the total purchase amount of ID POS and the total purchase amount for each cluster code, and stores the calculated product purchase rate (real value). In addition, the product purchase rate (theoretical value) predicted is also stored.

[Map information storage unit 24]
The map information storage unit 24 stores area map information corresponding to the cluster information, and the analysis result is displayed using the map information.

[Control unit 1]
As shown in FIG. 2, the control unit 1 reads the processing program stored in the storage unit 2, and performs function realization means such as a product purchase rate calculation unit 11, a product purchase rate prediction unit 12, and an analysis result display unit 13. Run.
[Product purchase rate calculation means 11]
The merchandise purchase rate calculation means 11 obtains the number of ID POS customers or the ID POS total purchase price corresponding to the cluster code from the ID POS information storage unit 22 for the cluster code of the clustered cluster information stored in the cluster information storage unit 21. The number of customers who purchase a specific product or the total purchase amount is read and aggregated for each cluster code (for each cluster), and the product purchase rate for each cluster code (number of product purchasers / ID POS customers or total purchase amount / ID POS total purchase amount) Is stored as a real value of the product purchase rate in the product purchase rate storage 23. A method for calculating the product purchase rate will be described later.

[Product purchase rate prediction means 12]
The merchandise purchase rate prediction means 12 determines whether to adopt the merchandise purchase rate for each cluster code according to a specific adoption standard for the calculated merchandise purchase rate (real value), and determines the product purchase rate. Using the factor score of the factor used in the cluster analysis as an explanatory variable and the product purchase rate adopted as the objective variable, a calculation formula (prediction model) for multiple regression analysis is generated for The product purchase rate (theoretical value) is predicted for all regions including the region corresponding to the cluster code for which the purchase rate is not adopted. A method for predicting the product purchase rate will be described later.

[Analysis result display means 13]
The analysis result display means 13 reads the map information stored in the map information storage unit 24 using the predicted product purchase rate (theoretical value) of the region, and displays the product purchase rate (theoretical value) on the map information. indicate. The analysis result display method will be described later.

[Interface part 3]
The interface unit 3 is an interface unit that is connected to the display device 4 and the input device 5, and is an interface unit that is also connected to a network.
The control unit 1 reads necessary information from the network via the interface unit 3 and stores it in the storage unit 2.

[Display device 4]
The display device 4 displays the information generated by the analysis result display means 13. In addition, you may display the information regarding cluster information, customer information, a goods purchase rate, etc. on the display apparatus 4. FIG.
[Input device 5]
The input device 5 is a device that inputs necessary instructions to the control unit 1.

[Processing content of this system]
Next, merchandise purchase rate calculation processing, merchandise purchase rate prediction processing, and analysis result display processing in this system will be described with reference to FIGS. 3 is a diagram showing a table of cluster code assignment for each administrative boundary, FIG. 4 is a diagram showing a table of product purchase rates, and FIG. 5 is a table of cluster codes and product purchase rates for each administrative boundary. FIG. 6 is a diagram showing a table in which the product purchase rate for each administrative boundary is predicted, and FIG. 7 is a diagram showing a display example of the analysis result.

[Product purchase rate calculation processing: FIGS. 3 and 4]
The product purchase rate calculation means 11 realized by the control unit 1 determines, for the area (area) for which the product purchase rate is calculated, the area of the administrative community included in the area (for example, town 1, town 2,...). 3 is selected and the numerical values (factor scores) of factors 1 to 9 when the region is subjected to cluster analysis are obtained from the cluster information storage unit 21, and the corresponding cluster code is assigned to each region to generate the table of FIG. To do.
Next, the product purchase rate calculating means 11 totals the number of ID POS customers and the number of product purchased customers for each cluster code of FIG. 3, calculates the product purchase rate (%), and sets it in FIG.
Further, the product purchase rate calculating means 11 may calculate the product purchase rate (%) by totaling the total ID POS purchase amount and the total purchase amount for each cluster code.

[Product purchase rate prediction processing: FIGS. 4, 5, and 6]
The product purchase rate prediction means 12 calculates the product purchase effective value (number of product purchase customers x real product purchase rate (%)) and sets it in the table of FIG. The product purchase rate is adopted for items that are less than the threshold, and the product purchase rate is not adopted. In FIG. 4, the threshold value is “500”.
The product purchase effective value may be calculated by the total purchase price × the actual product purchase rate (%).

  As shown in FIG. 5, the product purchase rate prediction means 12 assigns a cluster code to each administrative boundary, sets the product purchase rate (real value) adopted for the assigned cluster code, Factors 1 to 9 for each administrative boundary shown in Fig. 6 are also set. The product purchase rate corresponding to the rejected cluster code is blank.

And the goods purchase rate prediction means 12 produces | generates the calculation formula of a prediction model by multiple regression analysis by making the goods purchase rate (real value) of the employ | adopted cluster code into an objective variable, and making factors 1-9 into explanatory variables, Based on the calculation formula of multiple regression, the product purchase rate is calculated and predicted for all administrative regions including the region where the product purchase rate is not adopted. This predicted product purchase rate is set as the product purchase rate (theoretical value) in FIG.
Therefore, the product purchase rate (real value) is calculated for each cluster, and the product purchase rate (theoretical value) is calculated by the calculation formula of the prediction model using the explanatory variable (factor score) for each region. However, the theoretical value is more suitable for the prediction model and may not necessarily match the real value.

[Calculation formula for multiple regression analysis]
A method of obtaining a calculation formula (prediction model) for multiple regression analysis will be described.
The calculation formula of the multiple regression analysis generated by the multiple regression analysis process is simply expressed by the following formula, for example.
y = b0 + b1 .x1 + b2 .x2 + b3 .x3 +... + b9 .x9

  Here, “y” represents a potential such as a predicted value, “b0 to b9” is a coefficient, and “x1 to x9” is a factor score of factors 1 to 9. “Y” is an objective variable, and “x1 to x9” is an explanatory variable. “B0” is called “intercept” in the calculation formula, and is a contact point between the xy axis and the y axis. “B1 to b9” are coefficients for each factor.

For the cluster area where the product purchase rate is adopted, the adopted product purchase rate is the objective variable, and the factor scores of factors 1 to 9 for each region are used as explanatory variables x1 to x9. The coefficients b0 to b9 for all the regions are obtained by multiple regression analysis, and the calculation formula of the prediction model is generated.
Then, for the cluster regions where the product purchase rate is adopted or not adopted (all regions), the factor score of the region is substituted into the above formula to calculate the product purchase rate “y” of the region.
The product purchase rate predicted as described above is set in the table of FIG. 6 and stored in the product purchase rate storage unit 23 of the storage unit 2.

The above all areas are all areas in the area (area) for which the product purchase rate is calculated in FIG. 3. For example, when the area is “Tokyo”, the area “Sapporo” If it is analyzed and classified, the product purchase rate may be calculated and predicted using the explanatory variable of the factor score using the calculation formula of the multiple regression analysis.
That is, if a calculation formula (prediction model) of multiple regression analysis can be obtained in a specific area, the product purchase rate in the region of the whole country can be predicted using the prediction model.

[Analysis result display processing: FIG. 7]
The analysis result display means 13 reads the table predicting the product purchase rate (theoretical value) in FIG. 6 from the product purchase rate storage unit 23, reads the map information of the area including the region from the map information storage unit 24, and The product purchase rate for each region (theoretical value) is displayed on the map information. FIG. 7 shows an example in which regions where the product purchase rate is equal to or higher than a specific value are colored or shaded. In FIG. 7, the location information of the store from which the customer information has been acquired is displayed.

  In this system, the factor score used for cluster analysis is used to obtain the product purchase rate of a specific product for each region, so the purchase rate of a specific product in the region can be used as an index (index for a specific product) .

  Further, in this system, the analysis results differ depending on the type and form of the store, for example, convenience stores, supermarkets, drug stores, home centers, home appliance mass retailers, etc., so it is necessary to consider the interpretation of the results. For that purpose, it is better to describe the business form of the store in the analysis report.

[Effect of the embodiment]
According to this system, a specific product purchase rate in a region is clustered, and the number of ID POS customers and the number of product purchase customers or the total purchase amount of ID POS and the total purchase amount are totaled for each cluster to obtain the product purchase rate for each cluster. For a region in the product purchase rate cluster that uses more than a specific standard, the product purchase rate is the objective variable, and the factor score is the explanatory variable to obtain the calculation formula (prediction model) for multiple regression analysis For all regions analyzed, the product purchase rate of the region is calculated using the multiple regression analysis formula using the factor score for each region, so products using cluster analysis There is an effect that the product purchase rate can be appropriately predicted for all regions including the region where the purchase rate could not be obtained.

  In addition, according to this system, based on the calculation formula of the multiple regression analysis obtained in the analysis target area (area), the factor score used in the cluster analysis is used for the area in the area other than the area. Since the product purchase rate of a specific product is calculated and predicted, there is an effect that the product delivery rate can be predicted even in an area where customer information is not available in the store.

[Application Example 1]
In this system, a computer connected to a network is assumed. However, this is a system in which a client device that inputs an instruction and displays a processing result and a server device that realizes the main processing of this system are connected via the network. May be.
In this case, the cluster information storage unit 21, the ID POS information storage unit 22, the product purchase rate storage unit 23, and the map information storage unit 24 are made into a database (DB), and each DB is connected to the network to connect the server device and the data. Processing may be realized by exchanging.

  In the case of the above application example 1, the server device has a Web server function, and the user connects a personal computer to the network, and provides a report of the analysis result based on the product purchase rate in the database May be performed by the server device.

[Application Example 2]
In this system, regions are classified by administrative boundaries and meshes, but analysis units based on regions such as postal code boundaries, medical zones, school districts, etc. may be assumed.

[Application 3 / Purchase forecast analysis system using network: Fig. 8]
Next, a network-based purchase prediction analysis system will be described with reference to FIG. FIG. 8 is a schematic diagram of a network-based purchase prediction analysis system.
In the system of FIG. 8, the purchase prediction analysis server 30 is connected to a store center server 41 via a network 50.

  A cluster information storage unit 31, a product purchase rate storage unit 33, and a map information storage unit 34 are connected to the purchase prediction analysis server 30. Here, the cluster information storage unit 31, the product purchase rate storage unit 33, and the map information storage unit 34 correspond to the cluster information storage unit 21, the product purchase rate storage unit 23, and the map information storage unit 24 described in FIG. Similar data is stored.

An ID POS information storage unit 42 is connected to the store center server 41, and a client terminal (client PC) 40 is further connected via a network 43.
The store center server 41 collects the ID POS information of the store, stores it in the ID POS information storage unit 42, reads out the ID POS information from the ID POS information storage unit 42 according to an instruction from the client PC 40, and sends it to the purchase prediction analysis server 30. Send.
Here, the ID POS information storage unit 42 corresponds to the ID POS information storage unit 22 described with reference to FIG. 2, and stores similar data.
The client PC 40 is connected to the store center server 41, instructs to transmit the ID POS information to the purchase prediction analysis server 30, and displays a report of the analysis result transmitted from the purchase prediction analysis server 30.

[Operation of system in FIG. 8]
In the system shown in FIG. 8, the client PC 40 designates an area for obtaining the product purchase rate in the store center server 41, and the ID POS information (ID POS customer number, product purchase customer number or ID POS total purchase) for the product for which the product purchase rate is predicted. (Amount, total purchase amount) is read from the ID POS information storage unit 42 and sent to the purchase prediction analysis server 30.
The store center server 41 reads out the ID POS information from the ID POS information storage unit 42 and transmits it to the purchase prediction analysis server 30 together with information on the designated area.

The purchase prediction analysis server 30 calculates the product purchase rate (real value) by adding up the ID POS information from the store center server 41 to the cluster corresponding to the administrative boundary or mesh where the customer resides for each cluster. The purchase effective value is calculated, and the adoption or non-adoption of the product purchase rate (real value) is determined.
Then, the purchase prediction analysis server 30 generates the calculation formula of the prediction model by multiple regression analysis using the product purchase rate (real value) of the adopted cluster as the objective variable and the factors 1 to 9 as explanatory variables, and the multiple regression The product purchase rate is calculated and predicted for all administrative boundaries or mesh regions including the region where the product purchase rate is not adopted, and is obtained as a product purchase rate (theoretical value).

Further, the purchase prediction analysis server 30 generates a table in which the product purchase rate is predicted for each administrative boundary as shown in FIG. 6 for the area designated by the client PC 40 (for example, the area specified by the zip code list). And sent to the client PC 40 via the store center server 41.
The zip code list is a list in which a plurality of zip codes are described, and a plurality of administrative boundaries are specified by the zip codes.

The client PC 40 has an effect that the product purchase rate in the area can be appropriately predicted only by designating the ID POS information stored in the ID POS information storage unit 42 and the product purchase prediction area in the zip code list.
Here, since the store center server 41 manages ID POS information, the client PC 40 connected to the store center server 41 is assumed to be a store computer or a computer (terminal device) connected to the store computer. Server 30 operates as an application service provider (ASP).

  In the system of FIG. 8, the purchase prediction analysis server 30 obtains an index for measuring the degree of conformity of each area in the area with respect to the calculation formula (multiple regression model formula) of the prediction model of the multiple regression analysis, and stores the store center. It may be transmitted to the client PC 40 via the server 41.

  Here, the degree of fit to the multiple regression model formula is: multiple correlation coefficient: R, corrected R adjusted for it, determination coefficient: R square, corrected R square corrected, Durbin-Watson ratio, AIC ( Akaike information criterion), partial regression coefficient, standard error, standard partial regression coefficient, test of significance of partial regression coefficient: F value, test of significance of partial regression coefficient: t value, test of significance of partial regression coefficient : 95% confidence interval of P value and partial regression coefficient: lower limit, 95% confidence interval of partial regression coefficient: upper limit, correlation with objective variable: simple correlation, correlation with objective variable: partial correlation, multicollinearity Statistics: tolerance, multicollinearity statistics: VIF (dispersion expansion factor), etc., and these values serve as indices of the degree of conformity.

[Application Example 4 / Comparison of Product Purchase Rates: FIGS. 9 to 11]
Next, an example of a report that allows comparison of product purchase rates will be described with reference to FIGS.
[Product purchase rate comparison table: Fig. 9]
FIG. 9 shows a product purchase rate comparison table in which two products A and B are compared for each administrative boundary using the above method. FIG. 9 is a diagram showing a product purchase rate comparison table.
In the example of FIG. 9, the product purchase rates of the products A and B are compared for each administrative boundary in the same area.

Here, the product A and the product B may be competing products (competing company products), the same products before and after the sales promotion event, or the synchronous ratio (2012). The product purchase rate of the same product in the fiscal year and the fiscal year 2013 may be used.
Furthermore, a table comparing the product purchase rates of the same product or different products for different areas specified by different postal code lists may be created as a report. In this case, it is possible to create not only the ID POS information of the same store but also the ID POS information of different stores.
That is, the products A and B are products having different conditions such as type, period, and area.

[Spider chart comparing factor characteristics of product purchase: Fig. 10]
Next, a report example of a spider chart for comparing factor characteristics of product purchase will be described with reference to FIG. FIG. 10 is a spider chart for comparing factor characteristics of product purchases.
FIG. 10 is a spider chart in which the factor characteristics of the products A and B can be compared for each factor 1 to 9 based on the table of FIG. 6 created for the products A and B, respectively.
The products A and B include the same product at different times and the same or different products (products having different conditions) in different areas.

[Comparison table of upper and lower areas: Fig. 11]
In addition, in this system, when a multiple regression model formula is generated, a product purchase rate can be obtained for areas throughout the country based on cluster information. Therefore, the top 10 regions (Top10) and the bottom 10 regions (Bottom10) with the highest purchase rate of the product are calculated for the products A and B in the designated area (region) nationwide or designated and displayed side by side. This is the report example of FIG. FIG. 11 is a diagram showing a comparison table of upper and lower areas.
For the products A and B, the upper 10 and lower 10 regions (for example, areas b, d, and c) are shown in the designated area. You may make it include a goods purchase rate in the column of the calculated | required area | region.

  INDUSTRIAL APPLICABILITY The present invention is suitable for a purchase prediction analysis system that can predict a product purchase rate for each region including a region where a product purchase rate cannot be properly obtained, and a program thereof.

  DESCRIPTION OF SYMBOLS 1 ... Control part, 2 ... Memory | storage part, 3 ... Interface part, 4 ... Display apparatus, 5 ... Input device, 11 ... Product purchase rate calculating means, 12 ... Product Purchase rate prediction means, 13 ... analysis result display means, 21 ... cluster information storage unit, 22 ... ID POS information storage unit, 23 ... product purchase rate storage unit, 24 ... map information Storage unit 30 ... Purchase forecast analysis server 31 ... Cluster information storage unit 33 ... Product purchase rate storage unit 34 ... Map information storage unit 40 ... Client terminal (client PC) ), 41 ... store center server, 42 ... ID POS information storage unit, 43 ... network, 50 ... network

Claims (18)

A purchase prediction analysis system having an information processing device for predicting a regional product purchase rate,
A cluster information storage unit that stores cluster information that is cluster-analyzed and clustered for the region;
A customer information storage unit for storing customer information for a store in a specific area within the analysis target area;
A product purchase rate storage unit for storing the product purchase rate of the specific area,
As a function realizing unit realized by executing a processing program in the control unit of the information processing apparatus,
Based on the cluster information stored in the cluster information storage unit, from the customer information storage unit, the number of customers or the total purchase price of the customer and the number of product purchasers of a specific product or the total purchase of the product purchase customer for each cluster Product purchase rate calculation means for reading the amount of money, calculating the product purchase rate for each cluster, and storing it in the product purchase rate storage unit;
Adopting or not adopting the product purchase rate for each cluster using a specific threshold for the calculated product purchase rate, and purchasing the product for each region for the region corresponding to the cluster for which the product purchase rate is adopted Using the rate as the objective variable and the factor score used in the cluster analysis as the explanatory variable, generate a calculation model for the multiple regression analysis prediction model, and for the region corresponding to the cluster where the product purchase rate is adopted or not adopted, A purchase prediction analysis system comprising: a product purchase rate prediction unit that calculates and predicts a product purchase rate using a factor score of the region and the calculation formula, and stores the calculated product purchase rate in the product purchase rate storage unit. A purchase prediction analysis system having an information processing server for predicting a regional product purchase rate,
A cluster information storage unit that stores cluster information that is cluster-analyzed and clustered for the region;
A product purchase rate storage unit for storing the product purchase rate in a specific area within the analysis target area,
As a function realizing unit realized by executing a processing program in the control unit of the information processing apparatus,
Receives customer information for a store in a specific area within the analysis target area from a client via a network, and based on the cluster information stored in the cluster information storage unit, the number of customers for each cluster from the customer information Alternatively, the total purchase amount of the customer and the number of product purchasers of the specific product or the total purchase amount of the product purchase customer is read, the product purchase rate for each cluster is calculated, and the product purchase rate stored in the product purchase rate storage unit Computing means;
Adopting or not adopting the product purchase rate for each cluster using a specific threshold for the calculated product purchase rate, and purchasing the product for each region for the region corresponding to the cluster for which the product purchase rate is adopted Using the rate as the objective variable and the factor score used in the cluster analysis as the explanatory variable, generate a calculation model for the multiple regression analysis prediction model, and for the region corresponding to the cluster where the product purchase rate is adopted or not adopted, A purchase prediction analysis system comprising: a product purchase rate prediction unit that calculates and predicts a product purchase rate using a factor score of the region and the calculation formula, and stores the calculated product purchase rate in the product purchase rate storage unit. The product purchase rate calculation means reads the number of customers or the total purchase price of the customer and the number of product purchase customers of the specific product or the total purchase price of the product purchase customer for each cluster from the customer information. Divide the number by the number of customers, or divide the total purchase amount by the total purchase amount to calculate the product purchase rate for each cluster,
The product purchase rate prediction means obtains a product purchase effective value for each cluster by multiplying the number of the product purchase customers or the total purchase price and the product purchase rate, and the product purchase effective value is equal to or greater than a specific threshold value. 3. The purchase forecast analysis system according to claim 1, wherein the product purchase rate is adopted, and the product purchase rate is not adopted when the product purchase effective value is less than a specific threshold value. The product purchase rate storage unit also stores product purchase rates in regions other than the analysis target region,
The product purchase rate prediction means reads the cluster information for the region other than the analysis target region stored in the cluster information storage unit, and uses the factor score of the region and the generated multiple regression analysis prediction model calculation formula to purchase the product The purchase prediction analysis system according to any one of claims 1 to 3, wherein a rate is calculated and predicted, and stored in the product purchase rate storage unit. As a function realization means realized by executing a processing program in the control unit of the information processing apparatus,
It has an analysis result display means for reading the predicted product purchase rate in the region from the product purchase rate storage unit and displaying the product purchase rate or the characteristics of the factors used in the cluster analysis for products with different conditions The purchase forecast analysis system according to any one of claims 1 to 4 characterized by things. It has a map information storage unit that stores map information about the area,
As a function realization means realized by executing a processing program in the control unit of the information processing apparatus,
Read the predicted product purchase rate of the region from the product purchase rate storage unit, read the map information of the corresponding region from the map information storage unit, and display the corresponding region in the map information according to the product purchase rate for each region 5. The purchase forecast analysis system according to claim 1, further comprising analysis result display means for changing the form.   7. The purchase prediction analysis system according to claim 6, wherein the analysis result display means changes the display color of the corresponding region in the map information or the shade of the display color according to the product purchase rate.   7. The purchase forecast analysis system according to claim 6, wherein the analysis result display means displays a corresponding area in the map information in a special form for an area where the product purchase rate is a specific value or more.   The purchase prediction analysis system according to any one of claims 6 to 8, wherein the analysis result display means displays the location information of the store in a corresponding area in the map information. A purchase prediction analysis program that operates on an information processing device that predicts the product purchase rate in a region,
The information processing apparatus;
Cluster analysis is performed on the region stored in the cluster information storage unit, and the number of customers for a store in a specific region in the analysis target region for each cluster from the customer information storage unit based on the cluster information clustered or the customer's Product purchase rate calculation means for reading the total purchase price and the number of product purchasers of the specific product or the total purchase price of the product purchase customer, calculating the product purchase rate for each cluster, and storing it in the product purchase rate storage unit;
Adopting or not adopting the product purchase rate for each cluster using a specific threshold for the calculated product purchase rate, and purchasing the product for each region for the region corresponding to the cluster for which the product purchase rate is adopted Using the rate as the objective variable and the factor score used in the cluster analysis as the explanatory variable, generate a calculation model for the multiple regression analysis prediction model, and for the region corresponding to the cluster where the product purchase rate is adopted or not adopted, A purchase prediction analysis program that functions as a product purchase rate prediction unit that calculates and predicts a product purchase rate using a factor score of the region and the calculation formula, and stores it in the product purchase rate storage unit. A purchase prediction analysis program that operates on an information processing server that predicts a local product purchase rate,
The information processing server;
Receives customer information for a store in a specific area within the analysis target area from the client via the network, and performs cluster analysis on the area stored in the cluster information storage unit, and the customer is based on the clustered cluster information. For each cluster, the number of customers for the store in the specific area in the analysis target area or the total purchase price of the customer and the number of product purchasers of the specific product or the total purchase price of the product purchase customer is read A product purchase rate calculating means for calculating a product purchase rate for each product and storing it in a product purchase rate storage unit;
Adopting or not adopting the product purchase rate for each cluster using a specific threshold for the calculated product purchase rate, and purchasing the product for each region for the region corresponding to the cluster for which the product purchase rate is adopted Using the rate as the objective variable and the factor score used in the cluster analysis as the explanatory variable, generate a calculation model for the multiple regression analysis prediction model, and for the region corresponding to the cluster where the product purchase rate is adopted or not adopted, A purchase prediction analysis program that functions as a product purchase rate prediction unit that calculates and predicts a product purchase rate using a factor score of the region and the calculation formula, and stores it in the product purchase rate storage unit. The product purchase rate calculation means reads the number of customers or the total purchase price of the customer and the number of product purchase customers of the specific product or the total purchase price of the product purchase customer for each cluster from the customer information. Divide the number by the number of customers, or divide the total purchase amount by the total purchase amount to calculate the product purchase rate for each cluster,
The product purchase rate prediction means obtains a product purchase effective value for each cluster by multiplying the number of the product purchase customers or the total purchase price and the product purchase rate, and the product purchase effective value is equal to or greater than a specific threshold value. The purchase prediction analysis program according to claim 10 or 11, wherein a product purchase rate is adopted, and the product purchase rate is not adopted when the product purchase effective value is less than a specific threshold value.   The product purchase rate prediction means reads the cluster information for the region other than the analysis target region stored in the cluster information storage unit, and uses the factor score of the region and the generated multiple regression analysis prediction model calculation formula to purchase the product The purchase prediction analysis program according to any one of claims 10 to 12, wherein a rate is calculated and predicted, and stored in the product purchase rate storage unit as a product purchase rate in a region other than the analysis target region. Information processing device or information processing server
Functions as analysis result display means that reads the predicted product purchase rate of the region from the product purchase rate storage unit and displays the product purchase rate or the characteristics of the factors used for cluster analysis for products with different conditions The purchase forecast analysis system according to any one of claims 10 to 13 characterized by things. Information processing device or information processing server
Read the predicted product purchase rate of the region from the product purchase rate storage unit, read the map information of the corresponding region from the map information storage unit, and display the corresponding region in the map information according to the product purchase rate for each region The purchase forecast analysis program according to any one of claims 10 to 13, wherein the purchase forecast analysis program functions as an analysis result display means for changing an item.   16. The purchase prediction analysis program according to claim 15, wherein the analysis result display means changes the display color of the corresponding region in the map information or the shade of the display color according to the product purchase rate.   16. The purchase prediction analysis program according to claim 15, wherein the analysis result display means displays a corresponding area in the map information in a special form for an area where the product purchase rate is a specific value or more.   The purchase prediction analysis system according to any one of claims 15 to 17, wherein the analysis result display means displays the location information of the store in a corresponding area in the map information.

Images