Intelligent retail decision support system

Abstract

The present invention discloses an intelligent retail decision support system, including a data standardization module, a customer personality feature generation module, a customer grouping module, a model customization module and a target data generation module. By setting the customer personality feature generation module, the present invention helps users to extract some personalized features of customers and generate personalized feature portraits of customers, thereby realizing refined management of customers. By setting the customer grouping module, a large number of customers can be classified according to customer needs, which helps to retrieve standardized sales data in a targeted manner and obtain forecast data for a fixed customer group. By setting the model customization module, users can directly select the desired forecast model, which is convenient and quick. By linking with the customer grouping module, users can obtain specific target data for specific groups of people.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent retail, and in particular relates to an intelligent retail decision-making support system. Background Art

[0002] With the rapid development of the Internet and Internet of Things technologies, the concept of "new retail" has been proposed. New retail mainly uses computer technology to conduct a series of analyses on retail data to obtain information such as the best inventory level and the best selling price to help operators make decisions.

[0003] At present, smart retail software on the market is mainly divided into two categories: management and marketing. The management category focuses on the control of the marketing process, while the marketing category focuses on improving marketing capabilities.

[0004] The marketing smart retail software on the market mainly uses machine learning algorithms to predict target information. The disadvantage of this software is that it can only accurately predict the sales data of specific products and populations. Once the products or populations change, the accuracy will decrease. The reason is that sales behavior is based on people's subjective decisions. As time goes by, the characteristic classification of the population changes dynamically, and the types of goods are also constantly changing. If it is not possible to achieve refined processing of the population and goods, it will be difficult to obtain accurate prediction information. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide an intelligent retail decision-making support system in view of the deficiencies in the above-mentioned prior art, which can realize refined processing of people and commodities and obtain accurate prediction information.

[0006] To solve the above technical problems, the technical solution adopted by the present invention is: an intelligent retail decision-making support system, including a data standardization module, a customer personality feature generation module, a customer grouping module, a model customization module and a target data generation module;

[0007] The data standardization module is used to extract information from each piece of sales data input from the outside, and after adding, deleting, and modifying multiple pieces of information extracted from each piece of sales data, generate a piece of standardized sales data according to preset rules;

[0008] The customer personality feature generation module is used for allowing the user to set feature extraction rules, and generating a customer personalized feature portrait through standardized sales data according to the feature extraction rules set by the user;

[0009] The customer grouping module is used to group customers according to their personalized feature portraits to obtain the grouping feature portraits;

[0010] The model customization module is used for users to set model generation conditions, build intelligent models according to the model generation conditions, and save the built intelligent modules;

[0011] The target data generation module is used for users to select and save intelligent models, and to select standardized sales data of all customers in a group corresponding to at least one clustering feature portrait or personalized feature portraits of all customers in a group corresponding to at least one clustering feature portrait; and to use the selected standardized sales data or personalized feature portraits as input to the intelligent model to generate target data.

[0012] The above-mentioned intelligent retail decision-making support system, the data standardization module includes a data import unit, an information interception unit, an information marking unit, an information deletion unit, an information writing unit and a query unit;

[0013] The data import unit is used to import sales data from outside;

[0014] The information interception unit is used to intercept the sales data according to the interception rules set by the user;

[0015] The information marking unit is used for allowing users to mark information on sales data;

[0016] The information deletion unit is used for allowing the user to delete the intercepted information;

[0017] The information writing unit is used to write the marked information and the remaining information after deletion into the data template to generate standardized sales data;

[0018] The query unit is used for allowing users to query data templates.

[0019] In the above-mentioned intelligent retail decision-making support system, the customer personality feature generation module includes a feature selection unit and a feature extraction unit;

[0020] The feature selection unit is used for allowing the user to select features to be extracted;

[0021] The feature extraction unit is used to extract features selected by the feature selection unit and generate a personalized feature portrait of the customer based on the extraction results.

[0022] In the above-mentioned intelligent retail decision support system, the customer grouping module uses a clustering algorithm to group customers and obtain a grouping feature portrait.

[0023] In the above-mentioned intelligent retail decision-making support system, the model customization module includes a prediction target selection unit and a first time selection unit;

[0024] The prediction target selection unit is used for allowing the user to select a desired prediction target data type, and the target data types preset in the prediction target selection unit include repurchase rate and optimal price;

[0025] The first time selection unit is used for allowing the user to select a time point or time period for which the target data is to be predicted.

[0026] The above-mentioned intelligent retail decision-making support system, the target data generation module includes a model selection unit, a second time selection unit, a customer group selection unit and a target data generation unit;

[0027] The model selection unit is used for allowing the user to select a saved intelligent model;

[0028] The second time selection unit is used for allowing the user to select a time point or time period for extracting the required sample data; the sample data refers to the historical sales data or personalized feature portrait required for predicting the target data;

[0029] The customer group selection unit is used for allowing the user to select the customer group to be generated by the desired sample data;

[0030] The target data generation unit is used to extract the standardized sales data or personalized feature portraits of all customers in the group corresponding to the corresponding group feature portrait based on the selected customer group and the time information selected by the second time selection unit, and use the extracted standardized sales data or personalized feature portraits as sample data of the intelligent model selected by the user, and predict the target data within the time point or time period selected by the first time selection unit based on the sample data and the intelligent model.

[0031] The above-mentioned intelligent retail decision-making support system further includes the automatic purchasing module and the automatic pricing module;

[0032] The automatic purchase module is used to perform purchase operations according to target data and commodity inventory data;

[0033] The automatic pricing module is used to perform pricing operations according to the optimal price when the target data type is the optimal price.

[0034] Compared with the prior art, the present invention has the following advantages:

[0035] 1. The present invention sets a data standardization module, which can add, delete and modify multiple pieces of information in the sales data, and then generate a piece of standardized sales data according to preset rules, which is beneficial to subsequent data applications.

[0036] 2. The present invention, by setting up a customer personality feature generation module, helps users to extract some personalized features of customers and generate personalized feature portraits of customers, thereby achieving refined management of customers.

[0037] 3. The present invention can classify a large number of customers according to their needs by setting up a customer grouping module, which helps to retrieve standardized sales data in a targeted manner and obtain forecast data for a fixed customer group.

[0038] 4. The present invention sets up a model customization module, so that users can directly select the desired prediction model, which is convenient and quick. By linking with the customer grouping module, users can obtain specific target data for specific groups of people.

[0039] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a system module block diagram of the present invention.

[0041] Figure 2 A schematic diagram of a data template.

[0042] Figure 3 This is a unit block diagram of the data standardization module of the present invention.

[0043] Figure 4 It is a unit block diagram of the customer personality feature generation module of the present invention.

[0044] Figure 5 A unit block diagram of a customized module for the model of the present invention.

[0045] Figure 6 This is a unit block diagram of the target data generation module of the present invention. DETAILED DESCRIPTION

[0046] like Figure 1 As shown, an intelligent retail decision support system includes a data standardization module 100, a customer personality feature generation module 200, a customer grouping module 300, a model customization module 400 and a target data generation module 500.

[0047] like Figure 3 As shown, in this embodiment, the data standardization module 100 is used to extract information from each sales data input from the outside, and after adding, deleting and modifying the multiple information extracted from each sales data, generate a standardized sales data according to preset rules.

[0048] Specifically, the data standardization module 100 includes a data importing unit 101 , an information intercepting unit 102 , an information marking unit 103 , an information deleting unit 104 , an information writing unit 105 and a querying unit 106 .

[0049] The data import unit 101 is used to import sales data from the outside; the information contained in the sales data includes product information, consumer information, purchase time information, price information, profit information and inventory information; the sales data can be directly exported from the dealer's ERP system, but the sales data formats in different ERP systems are different, and the information contained is also very different, so after the sales data is imported, it is necessary to perform standardization operations.

[0050] The information interception unit 102 is used to intercept the sales data according to the interception rules set by the user; when the information interception unit 102 is actually used to perform the interception operation, the interception rules can be set for the sales data output by different ERP systems.

[0051] The information marking unit 103 is used for users to mark information on sales data; the information marking unit 103 here is an interactive unit, and users can mark new information on sales data. For example, assuming that a certain sales data is "Shovel quantity 1 Wang Dayuan purchased on May 1st", after seeing it, the user can mark the sales data with a "holiday" information to indicate that this sales data was generated on a holiday.

[0052] The information deletion unit 104 is used for the user to delete the intercepted information. The information deletion unit 104 is mainly used to delete some unnecessary information.

[0053] The information writing unit 105 is used to write the marked information and the remaining information after deletion into the data template to generate standardized sales data;

[0054] The query unit 106 is used for users to query data templates. Users can view the standardized sales data through data templates, and the standardized sales data format can also be realized by setting a list of data templates. Figure 2 The figure shows a schematic diagram of a data template.

[0055] like Figure 4 As shown, in this embodiment, the customer personality feature generation module 200 is used for the user to set feature extraction rules, and generates a customer personalized feature portrait through standardized sales data according to the feature extraction rules set by the user.

[0056] Specifically, the customer personality feature generation module 200 includes a feature selection unit 201 and a feature extraction unit 202; the feature selection unit 201 is used for allowing the user to select features to be extracted; the feature extraction unit 202 is used to extract features selected by the feature selection unit 201, and generate a customer personalized feature portrait based on the extraction results.

[0057] It should be noted that the features selected by the feature selection unit 201 are the information contained in the standardized sales data. Assuming that a user wants to obtain a shopping time feature portrait of a certain customer, the shopping time information in all the sales data of the customer within a period of time is extracted, and a time curve is generated as the customer's personalized feature portrait. The user can generate different customer personalized feature portraits according to actual needs.

[0058] In this embodiment, the customer grouping module 300 is used to group customers according to their personalized feature portraits to obtain group feature portraits.

[0059] Specifically, the customer grouping module 300 uses a clustering algorithm to group customers and obtain a grouping feature portrait. The classification algorithm may use a k-means algorithm or a hybrid algorithm of k-means and linkage clustering.

[0060] In this embodiment, the clustering operation is unsupervised cold start clustering. Specifically, the system performs unsupervised clustering based on the previously generated customer personalized feature portrait. The algorithm uses K-means combined with Linkage clustering to add newly defined cluster centers and cluster merging distances. The specific process is described as follows:

[0061] Step 1: Select the initial number of clusters k>100. Here, the algorithm first attempts to use the improved kmeans method (see step 2 below) to generate multiple small clusters, and then uses the Linkage clustering method to merge the small clusters (i.e., merge the cluster centers);

[0062] Step 2: Select k random points from the data as cluster centers, and then randomly select centers for each cluster. Here, the center of the data cluster group is calculated using the center value of the data point with the highest density, that is, the multidimensional vector average:

[0063] Here x j It is a data point that meets the data density requirement within a cluster.

[0064] In particular, users can select features in a customized manner according to their needs. For example, in the customer credit card usage business scenario, after the user generates the utilization rate, the utilization rate and age group information are added to the feature group to participate in the clustering. After unsupervised customer clustering, the system can provide comprehensive features for each group based on information summary. These comprehensive features can be standardized into index indices (1-10) to indicate the uniqueness of each cluster. For example, the mean age index of the first cluster is 2, the mean age index of the second cluster is 7, the most recent purchase amount index of the first cluster is 9, the mean age index of the second cluster is 5, and so on. For example, as shown in the following table:

[0065] Grouping Age Index Three-month purchase amount Six-month purchase amount A 3.1 8.2 10 B 7.5 2.5 1.2 C 5.2 5.3 7.8 D 9.3 4.1 2.4 E 2.9 9.1 5.3

[0066] like Figure 5 As shown, in this embodiment, the model customization module 400 is used for the user to set model generation conditions, build an intelligent model according to the model generation conditions, and save the built intelligent model.

[0067] Specifically, the model customization module 400 includes a prediction target selection unit 401 and a first time selection unit 402 .

[0068] The prediction target selection unit 401 is used for the user to select the target data type to be predicted. The target data types preset in the prediction target selection unit 401 include repurchase rate and optimal price. When the user selects the repurchase rate, the pre-trained repurchase rate model is used. The repurchase rate model is an existing mature technology model and is not described here. When the user selects the optimal price, it means that the profit maximization mathematical model is selected. In practice, the user can design different models according to the actual business scenario. In addition, in practice, the module customization module can be designed as an open source model here, and the user can embed the intelligent model he needs.

[0069] The first time selection unit 402 is used for the user to select the time point or time period for the target data to be predicted. The first time selection unit 402 is designed to help the user select the time range of the sales data input into the intelligent model. Because the amount of data in actual commodity sales is huge, and the fluctuation of sales data is greatly affected by factors such as government policies, promotions, and surrounding environment, it is not the case that the more data there is, the more accurate the prediction results of the intelligent model will be. Therefore, it is necessary to select appropriate data. The setting of the first time selection unit 402 allows users to have a choice of sales data in the time dimension.

[0070] When the user selects the target data type using the prediction target selection unit 401 and selects the time period or time point using the first time selection unit 402, a user-named intelligent model is generated.

[0071] like Figure 6 As shown, in this embodiment, the target data generation module 500 is used for allowing the user to select a saved intelligent model, and to select standardized sales data of all customers in a group corresponding to at least one clustering feature portrait or personalized feature portraits of all customers in a group corresponding to at least one clustering feature portrait; and to use the selected standardized sales data or personalized feature portrait as input of the intelligent model to generate target data.

[0072] Specifically, the target data generation module 500 includes a model selection unit 501, a second time selection unit 502, a customer group selection unit 503 and a target data generation unit 504; the model selection unit 501 is used for the user to select a saved intelligent model; the second time selection unit 502 is used for the user to select a time point or time period for extracting the required sample data; the sample data refers to the historical sales data or personalized feature portrait required for predicting the target data; the customer group selection unit 503 is used for the user to select the customer group generated by the required sample data; the target data generation unit 504 is used to extract the standardized sales data or personalized feature portraits of all customers in the group corresponding to the corresponding grouping feature portrait according to the selected customer group and the time information selected by the second time selection unit 502, and use the extracted standardized sales data or personalized feature portraits as the sample data of the intelligent model selected by the user, and predict the target data within the time point or time period selected by the first time selection unit 402 based on the sample data and the intelligent model.

[0073] In this embodiment, the automatic purchase module 600 and the automatic pricing module 700 are also included;

[0074] The automatic purchase module 600 is used to perform purchase operations according to target data and commodity inventory data; the automatic purchase module 600 communicates with the supplier sales system and can directly place orders;

[0075] The automatic pricing module 700 is used to perform pricing operations according to the optimal price when the target data type is the optimal price. The automatic pricing module 700 is connected to the store settlement system and will release pricing information 3-5 days in advance so that the staff can replace the product price tags.

[0076] The principle of the present invention is explained by taking the pre-purchase quantity of bread of a merchant as an example.

[0077] Assume that the user is a purchasing specialist for a supermarket chain.

[0078] The user imports and standardizes the sales data in the ERP system of each store through the data standardization module 100 in the present invention, and then uses the customer personality feature generation module 200 to generate a personality feature portrait of each customer who buys bread within a period of time, and then uses the customer grouping module 300 to group the customers who buy bread. The grouping is divided by the type of bread purchased. After grouping, the user can predict the purchase volume of bread for different customer groups. Specifically, the user selects the repurchase rate and the sales data in the first three months through the model customization module 400, and then selects the target group customers as the customer group that buys cream bread through the target data generation module 500, and predicts the purchase volume of cream bread every day in the next two weeks. At this time, the system inputs the cream bread purchase data of the customer group in the past three months as sales data into the pre-order rate model, calculates the pre-order rate of the group, and directly determines the purchase quantity by combining the current cream bread commodity inventory with the pre-order rate. The biggest feature of the present invention is that it allows users to make refined predictions on customers or customer groups, which facilitates user decision-making.

[0079] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any way. Any simple modification, change and equivalent structural change made to the above embodiment based on the technical essence of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. An intelligent retail decision support system, characterized by: It includes data standardization module, customer personality feature generation module, customer grouping module, model customization module and target data generation module; The data standardization module is used to extract information from each piece of sales data input from the outside, and after adding, deleting, and modifying multiple pieces of information extracted from each piece of sales data, generate a piece of standardized sales data according to preset rules; The customer personality feature generation module is used for allowing the user to set feature extraction rules, and generating a customer personalized feature portrait through standardized sales data according to the feature extraction rules set by the user; A customer grouping module is used to group customers according to their personalized feature portraits to obtain group feature portraits; the customer grouping module uses a clustering algorithm to group customers; A model customization module is used for allowing users to set model generation conditions, and to build an intelligent model according to the model generation conditions, and to save the built intelligent model; the model customization module includes a prediction target selection unit and a first time selection unit; the prediction target selection unit is used for allowing users to select the type of target data to be predicted, and the target data types preset in the prediction target selection unit include repurchase rate and optimal price, and the first time selection unit is used for allowing users to select the time point or time period for the target data to be predicted; A target data generation module is used for allowing a user to select a saved intelligent model, and to select standardized sales data of all customers in a group corresponding to at least one grouping feature portrait or personalized feature portraits of all customers in a group corresponding to at least one grouping feature portrait; and to use the selected standardized sales data or personalized feature portraits as inputs to the intelligent model to generate target data; The target data generation module includes a model selection unit, a second time selection unit, a customer group selection unit and a target data generation unit; The model selection unit is used for allowing the user to select a saved intelligent model; The second time selection unit is used for allowing the user to select a time point or time period for extracting the required sample data; the sample data refers to the historical sales data or personalized feature portrait required for predicting the target data; The customer group selection unit is used for allowing the user to select the customer group to be generated by the desired sample data; The target data generation unit is used to extract the standardized sales data or personalized feature portraits of all customers in the group corresponding to the corresponding group feature portrait based on the selected customer group and the time information selected by the second time selection unit, and use the extracted standardized sales data or personalized feature portraits as sample data of the intelligent model selected by the user, and predict the target data within the time point or time period selected by the first time selection unit based on the sample data and the intelligent model.

2. The intelligent retail decision support system according to claim 1, characterized in that: The data standardization module includes a data import unit, an information interception unit, an information marking unit, an information deletion unit, an information writing unit and a query unit; The data import unit is used to import sales data from outside; The information interception unit is used to intercept the sales data according to the interception rules set by the user; The information marking unit is used for allowing users to mark information on sales data; The information deletion unit is used for allowing the user to delete the intercepted information; The information writing unit is used to write the marked information and the remaining information after deletion into the data template to generate standardized sales data; The query unit is used for allowing users to query data templates.

3. The intelligent retail decision support system according to claim 1 or 2, characterized in that: The customer personality feature generation module includes a feature selection unit and a feature extraction unit; The feature selection unit is used for allowing the user to select features to be extracted; The feature extraction unit is used to extract features selected by the feature selection unit and generate a personalized feature portrait of the customer based on the extraction results.

4. The intelligent retail decision support system according to claim 1, characterized in that: It also includes an automatic stocking module and an automatic pricing module; The automatic purchase module is used to perform purchase operations according to target data and commodity inventory data; The automatic pricing module is used to perform pricing operations according to the optimal price when the target data type is the optimal price.

Images