JP5424989B2 - POS data analysis apparatus, method and program - Google Patents

Description

  The present invention relates to a POS data analysis apparatus, method and program, and in particular, in a POS management (point-of-sales information management) in which a customer's purchase history is managed, POS data analysis apparatus, method, and program for supporting discovery.

  POS systems have become popular, and it is now possible to record “what customers”, “what products”, “at which store”, “when”, “how many”, and “how much”. These POS systems can make ordering / inventory management more efficient.

  In addition, by improving the processing power of computers and databases, it has become possible to analyze popular products and the situation of side-selling and find correlation rules between products and attributes. Considering the obtained correlation rules, it is possible to improve the sales by changing the product shelf configuration in the store, changing the purchased product, or the like.

  However, the extraction of the above association rules has the following problems.

(1) A large number of association rules are extracted;
(2) there is a known association rule;
For the above (1), an Apriori algorithm (for example, see Non-Patent Document 1), an FP-growth algorithm (for example, see Non-Patent Document 2), and the like have been proposed. In addition, an algorithm that reduces the amount of calculation by combining the advantages of these algorithms (for example, see Non-Patent Document 3) and a method for shortening the calculation time by parallel processing (for example, see Non-Patent Document 4) are also proposed. Has been.

  In addition, a method of eliminating association rules that are not interesting has been proposed (for example, see Non-Patent Document 5). This method only evaluates the association rule by a statistical test and is not effective in solving the problem (2).

R. Agrawal and R. Srikant, "Fast Algorithms for Mining Association Rules," In Proceedings of the International Conference on Very Large Data Bases, pp. 487-499, 1994. J. Han, J. Pei, and Y. Yin, "Mining frequent patterns without candidate generation," In Proceedings of the ACM SIGMOD Conference on Management of Data, pp.1-12, 2000. Masao Mori and Noriyuki Abeshima, "Comparison and Proposal of Algorithms for Market Basket Analysis", The Operations Research Society of Japan, Abstracts of Autumn Meeting, pp.308-310, 2003. Eigo Iwahashi and Hayato Yamana, "Fast pattern extraction speedup by FP-growth parallelization", IPSJ Technical Report, Database System Technical Report, pp.327-333, 2003. Takeshi Fukuda and Shinichi Morishita, "Visualization of association rules", IEICE Technical Report, Data Engineering, pp.41-48, 1995.

  However, the methods of Non-Patent Documents 1 to 4 described above are intended to extract correlation rules at a high speed, and the problem that there is a known correlation rule of the above problem (2) cannot be solved.

  The present invention has been made in view of the above points. In POS data analysis, a POS analysis apparatus capable of extracting side-by-side information that seems to be less known from a large number of association rules, and An object is to provide a method and a program.

In order to solve the above-mentioned problem, the present invention (Claim 1) is a POS data analyzer,
POS data storage means for storing POS data;
Product genre management means for reading POS data from the POS data storage means and extracting product genre information from the POS data;
Commodity co-sale feature information generating means for generating co-sale feature information indicating the co-sale characteristics of each product of the POS data in the POS data storage means, and storing in the product co-sale feature information storage means,
Based on the product genre information extracted by the product genre management means, calculate the similarity between the products in the product parallel feature information storage means, cluster the products for each product genre based on the similarity, Product cluster analysis means for storing a clustering result together with the similarity in a product clustering result storage means;
By referring to the product clustering result storage unit, the number of products existing in each cluster is counted, and there is a cluster having the number of products for which the value obtained by the average number of products per one cluster and the standard deviation is out of a predetermined range. A characteristic product genre discovery support means for determining that the product genre has a large variation in products,
Have

The present invention (Claim 2 ) is a POS data analysis device,
POS data storage means for storing POS data;
Product genre management means for reading POS data from the POS data storage means and extracting product genre information from the POS data;
Commodity co-sale feature information generating means for generating co-sale feature information indicating the co-sale characteristics of each product of the POS data in the POS data storage means, and storing in the product co-sale feature information storage means,
Based on the product genre information extracted by the product genre management means, calculate the similarity between the products in the product parallel feature information storage means, cluster the products for each product genre based on the similarity, Product cluster analysis means for storing a clustering result together with the similarity in a product clustering result storage means;
Commodity genre sale feature information that refers to the sale information between products in the POS data storage means, generates product genre sale feature information including the feature quantity of the sale status of each product genre, and stores it in the product genre sale feature information storage means. Generating means;
Acquires the co-sale information from the product co-sale feature information storage unit, acquires the product genre of each product based on the co-sale information, refers to the product genre co-sale feature information storage unit based on the product genre, and the calculated value based on the feature quantity between the product category has a characteristic product discovery support unit determines the presence or absence of Heibai relationship, a.

The present invention (Claim 3 ) is a POS data analyzer,
POS data storage means for storing POS data;
Product genre management means for reading POS data from the POS data storage means and extracting product genre information from the POS data;
Commodity co-sale feature information generating means for generating co-sale feature information indicating the co-sale characteristics of each product of the POS data in the POS data storage means, and storing in the product co-sale feature information storage means,
Based on the product genre information extracted by the product genre management means, calculate the similarity between the products in the product parallel feature information storage means, cluster the products for each product genre based on the similarity, Product cluster analysis means for storing a clustering result together with the similarity in a product clustering result storage means;
Commodity genre sale feature information that refers to the sale information between products in the POS data storage means, generates product genre sale feature information including the feature quantity of the sale status of each product genre, and stores it in the product genre sale feature information storage means. Generating means;
By referring to the product clustering result storage unit, the number of products existing in each cluster is counted, and there is a cluster having the number of products for which the value obtained by the average number of products per one cluster and the standard deviation is out of a predetermined range. In this case, characteristic product genre discovery support means for extracting product genres with large variations in products,
Select a product out of the cluster with respect to the product genre extracted by the characteristic product genre discovery support means, refer to the POS data storage means based on the product, and acquire the co-sale information of the product The product genre of each product is acquired based on the product sales information, the product genre combined feature information storage unit is referred to based on the product genre, and calculated based on the feature quantity between the two product genres. the values, the characteristic items found support unit determines the presence or absence of Heibai relationship has.

The present invention (Claim 4 ) is a POS data analysis method,
In an apparatus having POS data storage means for storing POS data, product co-sale feature information storage unit, product co-sale feature information generation unit, product clustering result storage unit, product genre co-sale feature information storage unit,
A product genre management means for reading POS data from the POS data storage means and extracting product genre information from the POS data;
Commodity co-sale feature information generating means generates co-sale feature information indicating the co-sale characteristics of each product of the POS data of the POS data storage unit, and stores the product co-sale feature information information in the product co-sale feature information storage unit;
The product cluster analysis means calculates the similarity between the products in the product co-sale feature information storage means based on the product genre information extracted in the product genre management step, and each product genre based on the similarity A product cluster analysis step of clustering products into the product clustering results and storing the clustering results together with the similarity in the product clustering result storage means;
The characteristic product genre discovery support means refers to the product clustering result storage means, counts the number of products existing in each cluster, and the value obtained from the average number of products per cluster and the standard deviation is out of a predetermined range. A characteristic product genre discovery support step for determining that the product genre has a large product variation when there is a cluster having a product number of
I do.

The present invention (Claim 5 ) is a POS data analysis method,
In an apparatus having POS data storage means for storing POS data, product co-sale feature information storage unit, product co-sale feature information generation unit, product clustering result storage unit, product genre co-sale feature information storage unit,
A product genre management means for reading POS data from the POS data storage means and extracting product genre information from the POS data;
Commodity co-sale feature information generating means generates co-sale feature information indicating the co-sale characteristics of each product of the POS data of the POS data storage unit, and stores the product co-sale feature information information in the product co-sale feature information storage unit;
The product cluster analysis means calculates the similarity between the products in the product co-sale feature information storage means based on the product genre information extracted in the product genre management step, and each product genre based on the similarity A product cluster analysis step of clustering products into the product clustering results and storing the clustering results together with the similarity in the product clustering result storage means;
Is commodity genre Heibai characteristic information generating means, the reference to Heibai information between items POS data storage unit, generates a product genre Heibai characteristic information containing a characteristic parameter of Heibai status of each product category, product genre Heibai feature information A product genre sales feature information generation step stored in the storage means;
FEATURES merchandise discovery support means, said product Heibai acquires Heibai information from the characteristic information storing means, to obtain a product genre of each product on the basis of said combination sales information, the product genre Heibai features based on the product category referring to the information storage means, the calculated value based on the feature quantity between the two product genre, intends row and characteristic product discovery assistance determining whether, of Heibai relationship.

The present invention (Claim 6 ) is a POS data analysis method,
In an apparatus having POS data storage means for storing POS data, product co-sale feature information storage unit, product co-sale feature information generation unit, product clustering result storage unit, product genre co-sale feature information storage unit,
A product genre management means for reading POS data from the POS data storage means and extracting product genre information from the POS data;
Commodity co-sale feature information generating means generates co-sale feature information indicating the co-sale characteristics of each product of the POS data of the POS data storage unit, and stores the product co-sale feature information information in the product co-sale feature information storage unit;
The product cluster analysis means calculates the similarity between the products in the product co-sale feature information storage means based on the product genre information extracted in the product genre management step, and each product genre based on the similarity A product cluster analysis step of clustering products into the product clustering results and storing the clustering results together with the similarity in the product clustering result storage means;
Is commodity genre Heibai characteristic information generating means, the reference to Heibai information between items POS data storage unit, generates a product genre Heibai characteristic information containing a characteristic parameter of Heibai status of each product category, product genre Heibai feature information A product genre sales feature information generation step stored in the storage means;
FEATURES merchandise genre discovery support means, by referring to the commodity clustering result storage means, and counting the items that exist in each cluster, a range value determined is given by the average items and the standard deviation per cluster A characteristic product genre discovery support step for extracting a product genre having a large variation in products when there is a cluster having the number of products outside,
The characteristic product discovery support means selects a product out of the cluster for the product genre extracted in the characteristic product genre discovery support step, and refers to the POS data storage means based on the product, Acquiring the product co-sale information, obtaining the product genre of each product based on the product co-sale information, referring to the product genre co-sale feature information storage means based on the product genre, between the two product genres of the calculated value based on the feature amount, intends row and characteristic product discovery assistance determining whether, of Heibai relationship.

Further, the present invention (Claim 7 ) is a POS data analysis program for causing a computer to function as each means constituting the POS data analysis apparatus according to any one of Claims 1 to 3 .

  The present invention exists in a large amount by clustering products and presenting the hierarchical structure of the clustering to the user by defining the distance between the products to be smaller as the degree of concurrent sales is higher in accordance with the status of product sales. From the association rules, it becomes easy to extract side sale information that seems to be less known (unexpected).

Moreover, according to the present invention, by presenting the recommended product genre as an analysis target,
It is easy to extract unexpected sales information from a large number of association rules.

  Further, according to the present invention, it is possible to easily extract unexpected sales information from among a large number of correlation rules by presenting sales information that is considered to be characteristic sales.

  Furthermore, by narrowing down the analysis target to product genres with large variation in parallel sales and finding characteristic parallel sales information from the genre, it extracts unexpected simultaneous sales information from a large number of association rules at high speed and efficiently. It becomes easy.

It is a block diagram of the POS data analysis device in the 1st Embodiment of this invention. It is an example of the flow of the POS data analysis process in the 1st Embodiment of this invention. It is an example of the POS data stored in POS DB in the 1st Embodiment of this invention. It is an example of goods genre DB in the 1st Embodiment of this invention. It is an example of goods DB in a 1st embodiment of the present invention. It is an example (the 1) of the co-sale characteristic information in the 1st Embodiment of this invention. It is an example (the 2) of the co-sale characteristic information in the 1st Embodiment of this invention. It is an example of the sequence chart of the utilization scene in the 1st Embodiment of this invention. It is an example of the figure (the figure of a hierarchical structure) of the clustering result in the 1st Embodiment of this invention. It is a block diagram of the POS data analyzer in the 2nd Embodiment of this invention. It is a block diagram of the POS data analyzer in the 3rd Embodiment of this invention. It is an example of the process of the characteristic goods discovery assistance part in the 3rd Embodiment of this invention. It is a method of calculating co-sale information in the third embodiment of the present invention. It is a block diagram of the POS data analyzer in the 4th Embodiment of this invention.

  Embodiments of the present invention will be described below with reference to the drawings.

[First Embodiment]
The present invention is a POS data analysis device that supports the discovery of a characteristic parallel sale by clustering the sales situation of the commodity in each commodity genre and highlighting the characteristic commodity in the commodity genre. This is based on the assumption that “a product of the same product genre will have a similar side-by-side sales situation”, and is characterized by finding a product whose sales situation is significantly different from other products.

  FIG. 1 shows the configuration of a POS data analysis apparatus according to the first embodiment of the present invention.

  The POS data analysis device 100A includes a POS data management unit 110, a product genre management unit 120, a product management unit 130, a product side sale feature information generation unit 140, a product cluster analysis unit 150, a product cluster display unit 160, and a side sales status display unit 170. It consists of five databases (POS DB 101, product genre DB 102, product DB 103, product co-sale feature information DB 104, product cluster DB 105).

  The co-sale status display unit 160 and the product cluster display unit 170 are connected to the user interface management unit 11 (hereinafter referred to as “UI management unit”) of the user terminal 10.

  FIG. 2 shows an example of the flow of POS data analysis processing in the first embodiment of the present invention.

  Step 101) The POS data management unit 110 receives POS data input from a POS terminal of each store via a network or the like and records it in the POS DB 101. FIG. 3 shows an example of POS data stored in the POS DB 101. The POS data includes a receipt number, store type, sales date, day of the week, customer sex, customer age, product name, product code, unit price, sales quantity, sales amount, manufacturer, product genre name, and genre code.

  Step 102) The merchandise genre management unit 120 extracts merchandise genre information (information of merchandise genre itself not related to sales / both sales) included in the POS data, and records it in the merchandise genre DB 102. FIG. 4 shows an example of registration in the product genre DB 102. The product genre ID, genre code, and product genre name are registered in the product genre DB 102.

  Step 103) Similar to the product genre management unit 120, the product management unit 130 extracts product information included in the POS data (information of the product itself not related to sales / both sales) and records it in the product DB 103. FIG. 5 shows a registration example of the product DB 103. In the product DB 103, product ID, product code, product name, unit price, manufacturer, and product genre ID are registered.

  Step 104) The product co-sale feature information generation unit 140 analyzes the POS data to generate co-sale feature information indicating the feature of the co-sale status of each product, and records the co-sale feature information in the product co-sale feature information DB 104. As an example of the combined sale feature information, collection information of receipt numbers from which the product is purchased can be considered. For example, in the POS data of FIG. 3, FIG. 6 shows an example when a set of receipt numbers is used as the feature information for sales. From FIG. 6, it can be seen that the product with ID “3040” was purchased with three receipts with receipt numbers “162747”, “182432”, and “192334”. As another example of the side selling feature information, vector information indicating which products are sold together can be considered. An example in which the same information as FIG. 6 is expressed as vector information is shown in FIG.

  Step 105) The product cluster analysis unit 150 calculates the similarity (product sales) between the products based on the feature information acquired by accessing the product DB 103 and the product sales feature information DB 104, and uses the information to calculate the similarity. The products are clustered for each product genre, and the clustering result is recorded in the product cluster DB 105. When a set as shown in FIG. 6 is used as the side selling feature information, the similarity calculation includes a Simpson coefficient and a Jacquard coefficient. When vector information is used as shown in FIG. 7, the similarity can be calculated from the cosine similarity, the Pearson correlation coefficient, or the like. In addition, as a clustering method at this time, for example, an agglomerative hierarchical clustering method can be used.

  Next, a usage scene after the above processing is completed will be described.

  FIG. 8 is an example of a usage scene sequence chart according to the first embodiment of the present invention.

  A user of POS data analysis (hereinafter referred to as an analyst) accesses the UI management unit 11 of the user terminal 10 and requests display of a clustering result of a certain product genre. The UI management unit 11 transmits the requested product genre to the product cluster display unit 170 of the POS data analysis apparatus 100A (step 201).

  The product cluster display unit 170 accesses the product cluster DB 105, expresses the hierarchical structure, which is the clustering result of the requested product genre, as a diagram (step 202), and returns it to the UI management unit 11. The UI management unit 11 displays the received diagram on the analyst screen (step 203). At this time, the analyst can re-request the clustering result of another product genre. FIG. 9 shows an example of the clustering result.

  Here, the analyst can find a product out of the cluster based on the displayed clustering result. For example, in FIG. 9, the product I is a product that is out of the cluster. A product that is out of the cluster means that the sales situation is different from other products. In other words, this product sales situation is characteristic, and there is a possibility that it is an unexpected sales situation.

  The analyst who finds a product out of the cluster requests display of the co-sold product of the product (step 204). Upon receiving the request, the UI management unit 11 informs the co-sale status display unit 160 of the POS analyzer 100A of the requested product. The co-sale status display unit 160 accesses the product co-sale characteristic information DB 104 and the product DB 103, obtains the product information sold together, and returns it to the UI management unit 11 (step 205). The UI management unit 11 displays the received product information on the analyst screen (step 206).

  As described above, according to the present embodiment, by looking at the clustering result diagram shown in FIG. By confirming the sales situation of the discovered product, it is possible to find an unexpected sales situation.

[Second Embodiment]
FIG. 10 is a configuration diagram of a POS data analysis apparatus according to the second embodiment of the present invention. In the figure, the same components as those in FIG.

  A POS data analysis device 100B shown in FIG. 10 includes a characteristic product genre discovery support unit 210 in addition to the existing units in the POS data analysis device 100A in the first embodiment shown in FIG.

  The characteristic product genre discovery support unit 210 accesses the product cluster DB 105, extracts one or a plurality of product genres with large product variations, and transmits them to the UI management unit 11. The UI management unit 11 presents the received product genre to the analyst. Thereby, the analyst can efficiently find the product genre to be analyzed.

  In addition, the following procedure can be considered as an example of a method for examining whether or not a certain product genre has a large variation in products.

  1) A clustering result in a certain product genre is examined (a similarity degree set in advance is checked, and a result of clustering based on the similarity degree is examined).

  2) Count the number of products present in each cluster and calculate the average number of products μ per cluster and the standard deviation σ.

  3) Check whether there is a cluster having the number of products of “μ−3σ or less” or “μ + 3σ or more”.

  4) If there is a cluster corresponding to the above condition 3), it is determined that the corresponding product genre has a large variation in the product.

  Thus, the analyst can select the product genre to be analyzed from the product genres when the product variation is large, without searching for the product genre that he / she is interested in from zero.

  According to the present embodiment, in the usage scene in the first embodiment shown in FIG. 8, it is necessary to specify the product genre to be analyzed by the user. For example, although there are hundreds of merchandise genres in a convenience store, it is very difficult to specify the merchandise genre to be analyzed from these. What solves this problem is the presentation of the product genre recommended as the analysis target by the characteristic product genre discovery support unit 210 in the second embodiment.

[Third Embodiment]
FIG. 11 is a configuration diagram of a POS data analysis apparatus according to the third embodiment of the present invention.

  In the figure, the same components as those in FIG.

  In addition to the existing configuration of the POS data analysis apparatus 100A in the first embodiment shown in FIG. 1, the POS data analysis apparatus 100C shown in FIG. It has a support unit 310 and a product genre combined feature information DB 301.

  The product genre co-sale feature information generation unit 320 accesses the POS DB 101, refers to the co-sale information between products, examines the co-sale status between product genres, and displays the feature of the co-sale status of each product genre. Generate information. As an example of the merchandise genre sales feature information, collection information of receipt numbers from which the merchandise genre is purchased can be considered. As an example of the product genre sales feature information in this case, FIG. 6 can be considered as in the product sales feature information. As another example, vector information indicating the product genre sold together can be considered. An example of vector information is shown in FIG. The product genre combined feature information generation unit 320 records the generated feature information in the product genre combined feature information DB 301.

  The characteristic product discovery support unit 310 presents a commodity that seems to be a characteristic parallel sale, for example, by the following procedure. An image of the processing is shown in FIG.

  1) Access to the product co-sale feature information DB 104 and extract all co-sale information.

  2) The product genre DB 102 and the product DB 103 are accessed, and the product genre of each product in the co-sale information is checked.

  3) Based on the product genre acquired in 2), the product genre side sale feature information DB 104 is accessed to check whether there is a side-by-side sales relationship between the two product genres.

  4) If there is no side-by-side relationship, determine that the side-to-side relationship may be unexpected.

  5) Repeat steps 2) to 4) for all the co-sale information acquired in 1).

  In addition, an example of a method for extracting the product sales information / product genre sales information required in the above steps 1) and 3) will be described below. Here, it is assumed that a receipt set is recorded as the feature information. An image of the processing is shown in FIG.

  1) Select the first registered product in the combined product feature information DB 104.

  2) Refer to a preset support level threshold value and compare with the support level of the corresponding product.

    → If greater than the threshold value, proceed to process 3).

    → If it is smaller than the threshold value, return to process 1) and select the next product.

  3) Access the POS DB 101 and check all the products sold together with the product.

  4) Calculate the degree of co-sale (confidence) for the co-sold product and the significance of the co-sale. As the method of calculating the degree of parallel sales, various methods such as the Simpson coefficient and the Jacquard coefficient in the set similarity calculation method, and the cosine similarity can be considered. Significance probability calculation methods include chi-square method and Fisher's exact test.

  5) Refer to the preset sales degree and threshold of significance, and compare with the corresponding product.

    → If both values clear the threshold, extract the corresponding co-sale.

    → If either cannot clear the threshold, return to process 1) and select the next product.

  6) Return to process 1) and select the next product.

  In the usage scene in the first embodiment shown in FIG. 8, it is necessary to specify the product genre to be analyzed by the user. For example, although there are hundreds of merchandise genres in a convenience store, it is very difficult to specify the merchandise genre to be analyzed from these. With respect to this problem, according to the present embodiment, as shown in FIG. 12, it is possible to present unexpected side-by-side sales information (both side sales of “beer A-cake F” in case b in FIG. 12).

[Fourth Embodiment]
FIG. 14 is a configuration diagram of a POS data analysis apparatus according to the fourth embodiment of the present invention. In this figure, the same components as those in FIGS. 1, 10, and 11 are denoted by the same reference numerals, and description thereof is omitted.

  The POS data analysis device 100D in the present embodiment shown in FIG. 14 includes a characteristic product discovery support unit 310, in addition to the existing units in the POS data analysis device 100A in the first embodiment shown in FIG. The product genre sales feature information generation unit 320, the characteristic product genre discovery support unit 210, and the product genre sales feature information DB 301 are provided. That is, the configuration is a combination of the configurations of the POS data analysis devices 100B and 100C of the second embodiment and the third embodiment.

  In the third embodiment, the presence / absence of a parallel sales relationship between corresponding product genres is examined for all the parallel sales information between products. This seems to require a large amount of processing and processing time. Therefore, in the present embodiment, the third implementation is performed while speeding up the processing by reducing the concurrent sales information to be analyzed by using the characteristic product genre discovery support unit 210 shown in the second embodiment. The feature is that it helps to find more characteristic side-selling than the form of.

  The processing flow in the present embodiment is shown below. It is assumed that product genre combined feature information is stored in the product genre combined feature information DB 301 in advance by the product genre combined feature information generation unit 320 of the third embodiment described above.

  1) The characteristic product genre discovery support unit 210 accesses the product cluster DB 105 and extracts one or a plurality of product genres with large product variations (similar to the processing in the second embodiment).

  2) The characteristic product discovery support unit 310 accesses the product cluster DB 105 and selects a product out of the cluster for the extracted product genre (new processing content of the fourth embodiment).

  3) The characteristic product discovery support unit 310 accesses the POS DB 101 and refers to the co-sale information regarding the selected product (new processing content of the fourth embodiment).

  4) It is determined whether or not the corresponding co-sale information is characteristic (similar to the processing of FIG. 12 of the third embodiment).

  In addition, the operation | movement of each component in said 1st-4th embodiment is constructed | assembled as a program, installed in the computer utilized as a POS data analysis apparatus, made to run, or distribute | circulates via a network. Is possible.

  Further, the constructed program can be stored in a portable storage medium such as a hard disk, a flexible disk, or a CD-ROM, and can be installed or distributed in a computer.

  The present invention is not limited to the above-described embodiments, and various modifications and applications are possible within the scope of the claims.

10 User terminal 11 User interface management unit 100A, 100B, 100C POS data analyzer 101 POS DB
102 Product Genre DB
103 Product DB
104 Commodity sale feature information DB
105 Product Cluster DB
110 POS data management unit 120 product genre management unit 130 product management unit 140 product co-sale feature information generation unit 150 product cluster analysis unit 160 co-sale status display unit 170 product cluster display unit 210 characteristic product genre discovery support unit 301 product genre co-sale feature information DB
310 characteristic product discovery support unit 320 product genre sales feature information generation unit

Claims (7)

POS data analyzer,
POS data storage means for storing POS data;
Product genre management means for reading POS data from the POS data storage means and extracting product genre information from the POS data;
Commodity co-sale feature information generating means for generating co-sale feature information indicating the co-sale characteristics of each product of the POS data in the POS data storage means, and storing in the product co-sale feature information storage means,
Based on the product genre information extracted by the product genre management means, calculate the similarity between the products in the product parallel feature information storage means, cluster the products for each product genre based on the similarity, Product cluster analysis means for storing a clustering result together with the similarity in a product clustering result storage means;
By referring to the product clustering result storage unit, the number of products existing in each cluster is counted, and there is a cluster having the number of products for which the value obtained by the average number of products per one cluster and the standard deviation is out of a predetermined range. A characteristic product genre discovery support means for determining that the product genre has a large variation in product,
A POS data analysis device characterized by comprising: POS data analyzer,
POS data storage means for storing POS data;
Product genre management means for reading POS data from the POS data storage means and extracting product genre information from the POS data;
Commodity co-sale feature information generating means for generating co-sale feature information indicating the co-sale characteristics of each product of the POS data in the POS data storage means, and storing in the product co-sale feature information storage means,
Based on the product genre information extracted by the product genre management means, calculate the similarity between the products in the product parallel feature information storage means, cluster the products for each product genre based on the similarity, Product cluster analysis means for storing a clustering result together with the similarity in a product clustering result storage means;
Commodity genre sale feature information that refers to the sale information between products in the POS data storage means, generates product genre sale feature information including the feature quantity of the sale status of each product genre, and stores it in the product genre sale feature information storage means. Generating means;
Acquires the co-sale information from the product co-sale feature information storage unit, acquires the product genre of each product based on the co-sale information, refers to the product genre co-sale feature information storage unit based on the product genre, and A characteristic product discovery support means for determining whether or not there is a co-sale relationship based on a value calculated based on a feature quantity between product genres;
POS data analysis device, characterized by have a. POS data analyzer,
POS data storage means for storing POS data;
Product genre management means for reading POS data from the POS data storage means and extracting product genre information from the POS data;
Commodity co-sale feature information generating means for generating co-sale feature information indicating the co-sale characteristics of each product of the POS data in the POS data storage means, and storing in the product co-sale feature information storage means
Based on the product genre information extracted by the product genre management means, calculate the similarity between the products in the product parallel feature information storage means, cluster the products for each product genre based on the similarity, Product cluster analysis means for storing a clustering result together with the similarity in a product clustering result storage means;
Commodity genre sale feature information that refers to the sale information between products in the POS data storage means, generates product genre sale feature information including the feature quantity of the sale status of each product genre, and stores it in the product genre sale feature information storage means. Generating means;
By referring to the product clustering result storage unit, the number of products existing in each cluster is counted, and there is a cluster having the number of products for which the value obtained by the average number of products per one cluster and the standard deviation is out of a predetermined range. In this case, characteristic product genre discovery support means for extracting product genres with large variations in products,
Select a product out of the cluster with respect to the product genre extracted by the characteristic product genre discovery support means, refer to the POS data storage means based on the product, and acquire the co-sale information of the product The product genre of each product is acquired based on the product sales information, the product genre combined feature information storage unit is referred to based on the product genre, and calculated based on the feature quantity between the two product genres. Characteristic product discovery support means for determining the presence or absence of a co-sale relationship according to the value,
POS data analysis device, characterized by have a. POS data analysis method,
In an apparatus having POS data storage means for storing POS data, product co-sale feature information storage unit, product co-sale feature information generation unit, product clustering result storage unit, product genre co-sale feature information storage unit,
A product genre management means for reading POS data from the POS data storage means and extracting product genre information from the POS data;
Commodity co-sale feature information generating means generates co-sale feature information indicating the co-sale characteristics of each product of the POS data of the POS data storage unit, and stores the product co-sale feature information information in the product co-sale feature information storage unit;
The product cluster analysis means calculates the similarity between the products in the product co-sale feature information storage means based on the product genre information extracted in the product genre management step, and each product genre based on the similarity A product cluster analysis step of clustering products into the product clustering results and storing the clustering results together with the similarity in the product clustering result storage means;
The characteristic product genre discovery support means refers to the product clustering result storage means, counts the number of products existing in each cluster, and the value obtained from the average number of products per cluster and the standard deviation is out of a predetermined range. If there is a cluster having the number of products to be a characteristic product genre discovery support step for determining that the product genre has a large variation in products,
POS data analysis method characterized by POS data analysis method,
In an apparatus having POS data storage means for storing POS data, product co-sale feature information storage unit, product co-sale feature information generation unit, product clustering result storage unit, product genre co-sale feature information storage unit,
A product genre management means for reading POS data from the POS data storage means and extracting product genre information from the POS data;
Commodity co-sale feature information generating means generates co-sale feature information indicating the co-sale characteristics of each product of the POS data of the POS data storage unit, and stores the product co-sale feature information information in the product co-sale feature information storage unit;
The product cluster analysis means calculates the similarity between the products in the product co-sale feature information storage means based on the product genre information extracted in the product genre management step, and each product genre based on the similarity A product cluster analysis step of clustering products into the product clustering results and storing the clustering results together with the similarity in the product clustering result storage means;
Is commodity genre Heibai characteristic information generating means, the reference to Heibai information between items POS data storage unit, generates a product genre Heibai characteristic information containing a characteristic parameter of Heibai status of each product category, product genre Heibai feature information A product genre sales feature information generation step stored in the storage means;
FEATURES merchandise discovery support means, said product Heibai acquires Heibai information from the characteristic information storing means, to obtain a product genre of each product on the basis of said combination sales information, the product genre Heibai features based on the product category A characteristic product discovery support step of referring to the information storage means and determining the presence / absence of a co-sale relationship based on a value calculated based on a feature quantity between two product genres;
POS data analysis wherein the intends rows. POS data analysis method,
In an apparatus having POS data storage means for storing POS data, product co-sale feature information storage unit, product co-sale feature information generation unit, product clustering result storage unit, product genre co-sale feature information storage unit,
A product genre management means for reading POS data from the POS data storage means and extracting product genre information from the POS data;
Commodity co-sale feature information generating means generates co-sale feature information indicating the co-sale characteristics of each product of the POS data of the POS data storage unit, and stores the product co-sale feature information information in the product co-sale feature information storage unit;
The product cluster analysis means calculates the similarity between the products in the product co-sale feature information storage means based on the product genre information extracted in the product genre management step, and each product genre based on the similarity A product cluster analysis step of clustering products into the product clustering results and storing the clustering results together with the similarity in the product clustering result storage means;
Is commodity genre Heibai characteristic information generating means, the reference to Heibai information between items POS data storage unit, generates a product genre Heibai characteristic information containing a characteristic parameter of Heibai status of each product category, product genre Heibai feature information A product genre sales feature information generation step stored in the storage means;
FEATURES merchandise genre discovery support means, by referring to the commodity clustering result storage means, and counting the items that exist in each cluster, a range value determined is given by the average items and the standard deviation per cluster A characteristic product genre discovery support step for extracting a product genre having a large variation in products when there is a cluster having the number of products outside,
The characteristic product discovery support means selects a product out of the cluster for the product genre extracted in the characteristic product genre discovery support step, and refers to the POS data storage means based on the product, Acquiring the product co-sale information, obtaining the product genre of each product based on the product co-sale information, referring to the product genre co-sale feature information storage means based on the product genre, between the two product genres A characteristic product discovery support step for determining whether or not there is a co-sale relationship based on a value calculated based on the feature amount of
POS data analysis wherein the intends rows. The POS data analysis program for functioning a computer as each means which comprises the POS data analysis apparatus of any one of Claims 1 thru | or 3 .

Images