Gas station non-oil product category optimization method, device and equipment and storage medium

Abstract

The invention relates to the technical field of category optimization, and provides a gas station non-oil product category optimization method, device and equipment and a storage medium, and the method comprises the steps: carrying out the site clustering based on the site feature data and non-oil product sales data of a site set, and obtaining a plurality of site categories; correspondingly determining an initial category structure of each station category according to non-oil feature data and non-oil inventory data of each station under each station category; correspondingly creating a category optimization model of each site category according to the initial category structure of each site category and the non-oil feature data, the non-oil sales volume data and the category sales proportion of each site under each site category; and solving the category optimization model of each site category according to a heuristic algorithm, and correspondingly obtaining a recommended category structure of each site category. According to the embodiment of the invention, the user experience and shelf utilization rate of the gas station convenience store can be improved.

Description

Technical Field

[0001] The present specification relates to the technical field of category optimization, and in particular to a method, device, equipment and storage medium for optimizing non-oil product categories at gas stations. Background Art

[0002] Category optimization (or category planning) is a technical solution to optimize the categories of goods sold and the number of shelves that each category can occupy. Category optimization can help operators obtain higher profits and provide consumers with more product choices and purchase convenience.

[0003] In fact, the current category optimization of non-oil products sold in gas station convenience stores generally takes the high profit margin of non-oil products as the optimization target. However, although the profit margin of some products is relatively low, customers have the need to buy or use them frequently. Therefore, it is of great significance to reasonably optimize the category of non-oil products sold in gas station convenience stores in order to take into account the user experience and shelf utilization of gas station convenience stores. Summary of the invention

[0004] The purpose of the embodiments of this specification is to provide a method, device, equipment and storage medium for optimizing non-oil product categories at gas stations to improve the user experience and shelf utilization of gas station convenience stores.

[0005] To achieve the above objectives, on the one hand, the embodiments of this specification provide a method for optimizing non-oil product categories at a gas station, including:

[0006] Perform site clustering based on site feature data of site collection and non-oil product sales data to obtain multiple site categories;

[0007] According to the non-oil product characteristic data and non-oil product inventory data of each site under each site category, an initial category structure of each site category is correspondingly determined;

[0008] According to the initial category structure of each site category, as well as the non-oil product characteristic data, non-oil product sales data and category sales ratio of each site under each site category, a category optimization model for each site category is created accordingly;

[0009] The category optimization model of each of the site categories is solved according to the heuristic algorithm, and the recommended category structure of each of the site categories is obtained accordingly.

[0010] In the non-oil product category optimization method for gas stations of the embodiment of this specification, the site clustering based on the site feature data and non-oil product sales data of the site set includes:

[0011] Performing correlation analysis on the full feature set formed by the site feature data based on the site set and the non-oil product sales data within a specified period to remove redundant features and form a selected feature set;

[0012] determining a hierarchical level for the selected feature set;

[0013] The sites are clustered using the selected features at each level in descending order.

[0014] In the non-oil product category optimization method for a gas station according to an embodiment of this specification, the determining of the hierarchical level of the selected feature set includes:

[0015] Using the site features in the selected feature set as input variables, calling the mixed negative binomial distribution method to predict the non-oil product sales volume of the site, and determining the first-level selected features in the selected feature set according to the prediction results;

[0016] Using the site features other than the first-level selected features in the selected feature set as input variables, calling the mixed negative binomial distribution method to predict the non-oil product sales volume of the site, and determining the second-level selected features in the selected feature set according to the prediction results;

[0017] Using the site features other than the first-level selected features and the second-level selected features in the selected feature set as input variables, calling the mixed negative binomial distribution method to predict the non-oil product sales volume of the site, and determining the third-level selected features in the selected feature set according to the prediction results;

[0018] The process is performed in this way until the selected features of a specified level in the selected feature set are obtained.

[0019] In the non-oil product category optimization method for gas stations of the embodiment of this specification, the site clustering is performed using the selected features of each level in a hierarchical order from high to low, including:

[0020] When the first-level selected features are discrete features, hierarchical clustering is performed using the non-oil sales volume of the site to obtain the temporary class label of the site;

[0021] Calculating a confusion matrix of discrete feature labels and the temporary class labels; the discrete feature labels are feature labels of the first level selected features;

[0022] Performing hierarchical clustering on the discrete feature labels using the columns of the confusion matrix as features;

[0023] Determining the primary clustering labels of all sites in the site set according to the hierarchical clustering results of the discrete feature labels;

[0024] If the number of labels of the currently accumulated cluster labels does not reach the quantity threshold, calculating the intra-class variance of each site category in the first-level cluster labels;

[0025] Using the second-level selected features, clustering the sites under the site category with the largest intra-class variance in the first-level clustering labels to the next level, and obtaining the second-level clustering labels of all sites under the site category;

[0026] If the number of labels of the currently accumulated cluster labels does not reach the quantity threshold, calculating the intra-class variance of each site category in the secondary cluster labels;

[0027] Using the third-level selected features, clustering the sites under the site category with the largest intra-class variance in the second-level clustering labels to the next level, and obtaining the third-level clustering labels of all sites under the site category;

[0028] This process is repeated in this way until the number of currently accumulated cluster labels reaches the threshold.

[0029] In the non-oil product category optimization method for gas stations of the embodiment of this specification, the site clustering is performed using the selected features of each level in a hierarchical order from high to low, including:

[0030] When the first-level selected feature is a continuous feature, the site non-oil sales hierarchical clustering is used to obtain the site temporary class label;

[0031] The distribution of continuous feature values ​​of sites clustered into the same category was estimated, and multiple normal distributions were obtained;

[0032] Arrange the multiple normal distributions in order, and calculate the intersection of density functions of every two adjacent normal distributions;

[0033] Dividing the first-level selected features into multiple intervals according to the intersection points, and clustering the sites in the same interval into one category, to obtain the first-level clustering labels of all sites in the site set;

[0034] If the number of labels of the currently accumulated cluster labels does not reach the quantity threshold, calculating the intra-class variance of each site category in the first-level cluster labels;

[0035] Using the second-level selected features, clustering the sites under the site category with the largest intra-class variance in the first-level clustering labels to the next level, and obtaining the second-level clustering labels of all sites under the site category;

[0036] If the number of labels of the currently accumulated cluster labels does not reach the quantity threshold, calculating the intra-class variance of each site category in the secondary cluster labels;

[0037] Using the third-level selected features, clustering the sites under the site category with the largest intra-class variance in the second-level clustering labels to the next level, and obtaining the third-level clustering labels of all sites under the site category;

[0038] This process is repeated in this way until the number of currently accumulated cluster labels reaches the threshold.

[0039] In the non-oil product category optimization method for gas stations of the embodiment of this specification, the initial category structure of each of the station categories is determined according to the non-oil product feature data and non-oil product inventory data of each station under each of the station categories, including:

[0040] For each site category, do the following:

[0041] Determine the commodity number vector of all sites under the site category, and take the maximum value in the commodity number vector as the maximum number of commodities for sale under the site category;

[0042] Based on the non-oil inventory data of all sites under this site category, construct the commodity sales information matrix of this site category;

[0043] Based on the non-oil product characteristic data of all sites under this site category, the category matrix, category brand matrix and category price band matrix of this site category at a given category granularity are constructed respectively;

[0044] The commodity sale information matrix is ​​multiplied by the category matrix to obtain a commodity number matrix of each category sold at each site under the site category; the rows of the commodity number matrix of each category are sites and the columns are commodity categories; and according to the minimum value of each column in the commodity number matrix of each category, a commodity number matrix of the least sold category of the site category is formed as the first commodity number matrix;

[0045] The commodity sales information matrix is ​​multiplied by the category brand matrix to obtain a commodity number matrix of various category brands sold at each site under the site category; the behavior of the commodity number matrix of each category brand is the site and the column is the category brand; and according to the minimum value of each column in the commodity number matrix of each category brand, a commodity number matrix of the least sold category brand of the site category is formed as the second commodity number matrix;

[0046] The commodity sale information matrix is ​​multiplied by the category price band matrix to obtain a commodity number matrix of each category price band sold at each site under the site category; the matrix of commodity number of each category price band is composed of sites and columns as category price bands; and according to the minimum value of each column in the matrix of commodity number of each category price band, a commodity number matrix of the least sold category price band of the site category is formed as the third commodity number matrix;

[0047] The maximum number of sold commodities under the site category, the first commodity number matrix, the second commodity number matrix and the third commodity number matrix are taken together as the initial category structure of the site category.

[0048] In the non-oil product category optimization method for gas stations of the embodiment of this specification, the general expression of the category optimization model for each of the station categories is:

[0049]

[0050] The corresponding constraints are:

[0051]

[0052] Among them, max is the maximum value objective function, x kj is whether the product j of the k-th site is sold, α is the hyperparameter that controls the weight of profit and sales, m is the number of products, n is the number of sites, K is, M j is the profit of commodity j, S ij is the demand for commodity j in the planning period at site i, T ik is whether site i belongs to the kth category site, p j is the proportion of sites where product j is sold at least in the system, For p j The product of n is rounded up, R k is the maximum number of commodities sold on the k-th category site, is whether product j belongs to the lth category, is the minimum number of commodities sold under the lth category in the kth category site, Is whether product j belongs to the lth category brand, is the minimum number of products sold under the brand of the lth category in the kth category site, Is whether product j belongs to the lth category price band, is the minimum number of commodities sold in the price band of the lth category in the kth category site, C ij is the purchase cost of product j in site i, G is the lower bound of gross profit rate of return, V ij is the forecast inventory for the planning period of item j in site i.

[0053] On the other hand, the embodiment of this specification also provides a non-oil product category optimization device for a gas station, including:

[0054] A site clustering module is used to cluster sites based on site feature data and non-oil product sales data of a site collection to obtain multiple site categories;

[0055] A structure determination module, used for determining the initial category structure of each site category according to the non-oil product characteristic data and non-oil product inventory data of each site under each site category;

[0056] A model creation module, for creating a category optimization model for each of the site categories according to the initial category structure of each of the site categories, and the non-oil product feature data, non-oil product sales data and category sales ratio of each site under each of the site categories;

[0057] The model solving module is used to solve the category optimization model of each of the site categories according to the heuristic algorithm, and obtain the recommended category structure of each of the site categories.

[0058] On the other hand, an embodiment of the present specification further provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein when the computer program is executed by the processor, the instructions of the above method are executed.

[0059] On the other hand, an embodiment of the present specification further provides a computer storage medium on which a computer program is stored. When the computer program is executed by a processor of a computer device, the instructions of the above method are executed.

[0060] On the other hand, an embodiment of the present specification further provides a computer program product, wherein the computer program product includes a computer program, and when the computer program is executed by a processor of a computer device, the computer program executes instructions of the above method.

[0061] It can be seen from the technical solutions provided in the above embodiments of this specification that, in the embodiments of this specification, the site clustering is first performed based on the site characteristic data and non-oil sales data of the site set; then the initial category structure is learned based on the non-oil characteristic data and non-oil inventory data of each site under each site category; and the category optimization model of each site category is created according to the initial category structure of each site category, as well as the non-oil characteristic data, non-oil sales data and category sales ratio of each site under each site category; the recommended category structure of each site category is obtained by solving the category optimization model, and the subsequent relevant personnel can perform site category configuration based on this, that is, perform site category configuration for all sites under the site category according to the recommended category structure of each site category. Since multiple influencing factors such as site characteristic data, non-oil sales data, initial category structure, non-oil inventory data, category sales ratio, etc. are considered in the category optimization process, the recommended category structure of each site category obtained after optimization is more accurate and reasonable, thereby improving the user experience and shelf utilization rate of gas station convenience stores, and reducing or avoiding the waste of logistics resources due to unreasonable category structure configuration of the site. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art description. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. In the drawings:

[0063] Figure 1 A schematic diagram showing an application environment for optimization of non-oil product categories at gas stations in some embodiments of this specification;

[0064] Figure 2 A flowchart of a method for optimizing non-oil product categories at a gas station in some embodiments of this specification is shown;

[0065] Figure 3 Shows Figure 2 A flow chart of performing site clustering based on site characteristic data and non-oil product sales data of a site set in the method shown;

[0066] Figure 4 Shows Figure 3 A flow chart for determining a hierarchical level for a selected feature set in the method shown;

[0067] Figure 5 Shows Figure 3 A flowchart of a method for clustering sites using selected features of each level in a hierarchical order from high to low;

[0068] Figure 6 Shows Figure 3 A flowchart of another method for clustering sites using selected features at each level in a hierarchical order from high to low in the illustrated method;

[0069] Figure 7 A schematic diagram showing clustering results of a site set in an exemplary embodiment of this specification is shown;

[0070] Figure 8 Shows Figure 2 A flowchart of the method shown, in which the initial category structure of each site category is determined according to the non-oil product characteristic data and the non-oil product inventory data of each site under each site category;

[0071] Fig. 9 A schematic diagram showing the dynamic changes in the inventory of goods from the current replenishment decision to the next arrival of goods in an exemplary embodiment of this specification is shown;

[0072] Fig.10 A structural block diagram of the device in some embodiments of this specification is shown;

[0073] Fig.11 The structure block diagram of the computer device in some embodiments of the present specification is shown.

[0074] [Description of Reference Numerals]

[0075] 10. Client;

[0076] 20. Server;

[0077] 101. Site clustering module;

[0078] 102. Structure determination module;

[0079] 103. Model creation module;

[0080] 104. Model solving module;

[0081] 1102. Computer equipment;

[0082] 1104. Processor;

[0083] 1106. Memory;

[0084] 1108. Driving mechanism;

[0085] 1110, input / output interface;

[0086] 1112. Input device;

[0087] 1114. Output device;

[0088] 1116. Presentation equipment;

[0089] 1118. Graphical user interface;

[0090] 1120, network interface;

[0091] 1122. Communication link;

[0092] 1124. Communication bus. DETAILED DESCRIPTION

[0093] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0094] Gas stations usually have convenience stores that mainly sell non-oil products (hereinafter referred to as non-oil products). In the embodiments of this specification, these convenience stores are referred to as sites (one convenience store is one site). In view of the problems of low user experience and low shelf utilization faced by category optimization of traditional sites, the embodiments of this specification provide an improved decision-making solution for optimizing non-oil product categories at gas stations.

[0095] Figure 1 , a schematic diagram of the application environment for optimizing the non-oil product categories of gas stations in some embodiments of the specification is shown; the application environment includes a client 10 and a server 20. The server 20 can perform site clustering based on the site feature data of the site set and the non-oil product sales data to obtain multiple site categories; determine the initial category structure of each site category according to the non-oil product feature data and non-oil product inventory data of each site under each site category; create a category optimization model for each site category according to the initial category structure of each site category, as well as the non-oil product feature data, non-oil product sales data and category sales ratio of each site under each site category; solve the category optimization model of each site category according to the heuristic algorithm, obtain the recommended category structure of each site category, and output the recommended category structure of each site category to the client 10 for display, so that relevant personnel can configure site categories accordingly, so as to achieve the purpose of improving the user experience and shelf utilization rate of gas station convenience stores.

[0096] In some embodiments of the present specification, the server 20 may be an electronic device with computing and network interaction functions; or may be software running in the electronic device and providing business logic for data processing and network interaction.

[0097] In some embodiments of the present specification, the client 10 may be a self-service terminal device, a mobile terminal (i.e., a smart phone), a display, a desktop computer, a tablet computer, a laptop computer, a digital assistant, or a smart wearable device, etc. Among them, the smart wearable device may include a smart bracelet, a smart watch, smart glasses, or a smart helmet, etc. Of course, the client 10 is not limited to the above-mentioned electronic devices with a certain entity, and it may also be software running in the above-mentioned electronic devices.

[0098] In addition, it should be noted that Figure 1 What is shown is only an application environment provided by this specification. In actual application, there may be multiple clients 10 and multiple servers 20, and this specification does not impose any limitation.

[0099] This specification embodiment provides a non-oil product category optimization method for a gas station, which can be applied to the above-mentioned service end side, referring to Figure 2As shown, in some embodiments of this specification, the non-oil product category optimization method for a gas station may include the following steps:

[0100] Step 201: Perform site clustering based on site feature data and non-oil product sales data of a site set to obtain multiple site categories;

[0101] Step 202: According to the non-oil product characteristic data and non-oil product inventory data of each site under each site category, an initial category structure of each site category is correspondingly determined;

[0102] Step 203: creating a category optimization model for each site category according to the initial category structure of each site category, and the non-oil product feature data, non-oil product sales data and category sales ratio of each site under each site category;

[0103] Step 204: Solve the category optimization model of each of the site categories according to the heuristic algorithm, and obtain the recommended category structure of each of the site categories.

[0104] In the embodiment of this specification, the site clustering is first performed based on the site characteristic data and non-oil sales data of the site set; then the initial category structure is learned based on the non-oil characteristic data and non-oil inventory data of each site under each site category; and the category optimization model of each site category is created according to the initial category structure of each site category, as well as the non-oil characteristic data, non-oil sales data and category sales ratio of each site under each site category; the recommended category structure of each site category is obtained by solving the category optimization model, and the subsequent relevant personnel can perform site category configuration based on this, that is, perform site category configuration for all sites under the site category according to the recommended category structure of each site category. Since multiple influencing factors such as site characteristic data, non-oil sales data, initial category structure, non-oil inventory data, category sales ratio, etc. are considered in the category optimization process, the recommended category structure of each site category obtained after optimization is more accurate and reasonable, thereby improving the user experience and shelf utilization rate of gas station convenience stores, and reducing or avoiding the waste of logistics resources due to unreasonable category structure configuration of the site.

[0105] Although different sites have different attributes (such as different distribution locations, different regular consumer groups, etc.), many sites have similar attributes (such as similar distribution locations, similar consumer groups, etc.) in large areas (such as the Northeast, Northwest, South China, and Jiangnan). Therefore, in order to take into account the accuracy and efficiency of site non-oil category optimization, all sites can be clustered to perform category optimization for each site class.

[0106] refer to Figure 3As shown, in some embodiments of the present specification, performing site clustering based on site feature data and non-oil product sales data of a site set may include the following steps:

[0107] Step 301: perform correlation analysis on a full feature set formed by site feature data based on a site set and non-oil product sales data within a specified period of time to remove redundant features and form a selected feature set.

[0108] A site collection is a site collection within a specified range, wherein the specified range can be customized as needed. For example, in an exemplary embodiment, the specified range can be one or more provincial administrative regions; in another exemplary embodiment, the specified range can also be a prefecture-level administrative region.

[0109] Site feature data refers to data used to characterize the attribute characteristics of the site itself. For example, geographical location, floor area, business hall area, parking lot area, type of gas station, etc. Among them, the type of gas station can include Class I gas stations (such as gas stations in municipalities, provincial capitals, and prefecture-level cities), Class II gas stations (such as gas stations on highways and ring roads above the prefecture level), Class III gas stations (such as gas stations in county-level cities, county towns, and national highways), and Class IV gas stations (such as gas stations on county and township roads, towns, water, and rural areas).

[0110] The non-oil sales data refers to the non-oil sales of a site within a specified time range. For example, in an exemplary embodiment, the non-oil sales data may be the non-oil sales of a site in the most recent quarter, the most recent year, or the most recent three years.

[0111] The site feature data based on the site set and the non-oil product sales can derive a large number of feature indicators, but in fact there may be features with strong feature correlation, resulting in feature redundancy. Therefore, it is necessary to perform correlation analysis on the full feature set formed by the site feature data based on the site set and the non-oil product sales data within a specified period of time to remove the redundant features and form a selected feature set.

[0112] In some embodiments of the present specification, for discrete features, correlation analysis can be implemented by methods such as chi-square test, information gain and information gain rate, etc. For continuous features, correlation analysis can be implemented by methods such as covariance and linear correlation coefficient, etc.

[0113] The selected feature set is the features retained after removing redundancy. The selected feature set may include both discrete features (such as geographic location, type of gas station, etc.) and continuous features (such as sales volume of non-oil products in different dimensions, floor space, business area, etc.).

[0114] In addition, since the oil sales of a gas station can reflect the passenger flow of the station to a certain extent, and the passenger flow will affect the sales of non-oil products at the station, etc. Therefore, in other embodiments of this specification, when performing station clustering, oil sales data (i.e., oil sales data of a station within a specified time range) may also be considered. For example, in an exemplary embodiment, the feature set considering oil sales data may be as shown in Table 1 below.

[0115] Table 1

[0116]

[0117]

[0118] Step 302: Determine the hierarchical level of the selected feature set.

[0119] In some embodiments of the present specification, a concept similar to a decision tree may be used to hierarchically divide sites, and on this basis, site clustering may be performed.

[0120] Step 303: Cluster the sites using the selected features of each level in descending order.

[0121] refer to Figure 4 As shown, in some embodiments of the present specification, determining the hierarchical level of the selected feature set may include the following steps:

[0122] Step 401: Using the site features in the selected feature set as input variables, calling the Compound Negative binomial distribution method to predict the non-oil sales volume of the site, and determining the first-level selected features in the selected feature set according to the prediction results.

[0123] In the embodiments of the present specification, the site characteristics in the selected feature set are used as input variables, and the mixed negative binomial distribution method is used to predict the non-oil product sales of the site under different input site characteristics; by comparing the non-oil product demand of the site with the actual sales of non-oil products, a selected feature that contributes the most to the demand can be determined, thereby forming the first-level selected features in the selected feature set.

[0124] Step 402: using the site features in the selected feature set except the first-level selected features as input variables, calling the mixed negative binomial distribution method to predict the non-oil sales volume of the site, and determining the second-level selected features in the selected feature set based on the prediction results.

[0125] Step 403: using the site features in the selected feature set except the first-level selected features and the second-level selected features as input variables, calling the mixed negative binomial distribution method to predict the non-oil sales volume of the site, and determining the third-level selected features in the selected feature set based on the prediction results.

[0126] Step 404, and so on, are performed until the selected features of a specified level in the selected feature set are obtained.

[0127] The designated level can be customized as needed. For example, if the designated level is level 2, the hierarchical level processing of the selected feature set is terminated after step 402 is performed. For another example, if the designated level is level 4, the hierarchical level processing of the selected feature set is terminated after the fourth level selected feature in the selected feature set is obtained.

[0128] In the hierarchical levels of the selected feature set, the first level selected features have the highest importance level, and the importance of the selected features of the subsequent levels decreases in turn. Therefore, when clustering, the selected features of each level can be used to cluster sites in a hierarchical order from high to low. In order to improve clustering accuracy, different processing methods can be used for hierarchical levels of different variable types.

[0129] refer to Figure 5 As shown, in some embodiments of the present specification, when the hierarchical selected features are discrete features, clustering sites using the selected features of each level in a hierarchical order from high to low may include the following steps:

[0130] Step 501: When the first-level selected feature is a discrete feature, hierarchical clustering is performed using the site non-oil product sales volume to obtain a temporary class label for the site.

[0131] The first-level selected features are discrete features, which means that all the selected features at the first level are discrete features. Hierarchical clustering is performed using the non-oil sales volume of the site (i.e., clustering is performed based on the non-oil sales volume of the site within a specified time range) to obtain temporary site class labels. If K classes are clustered (K is a positive integer), there are K temporary site class labels.

[0132] Step 502: Calculate the confusion matrix of the discrete feature labels and the temporary class labels; the discrete feature labels here refer to the feature labels of the first-level selected features.

[0133] A discrete feature label is a label used to uniquely identify a discrete feature.

[0134] Step 503: Using the columns of the confusion matrix as features, hierarchical clustering is performed on the discrete feature labels.

[0135] The confusion matrix, also known as the error matrix, is a standard visual format for evaluating prediction accuracy, and is generally represented by a matrix of n rows and n columns. In the embodiments of this specification, each column of the confusion matrix represents the predicted category of non-oil sales, and each row represents the actual category of non-oil sales. Therefore, by using the columns of the confusion matrix as features to perform hierarchical clustering on the discrete feature labels again, multiple categories (for example, clustered into K categories) can be obtained.

[0136] Step 504: Determine the primary clustering labels of all sites in the site set according to the hierarchical clustering results of the discrete feature labels.

[0137] In the obtained clustering results, the sites corresponding to the discrete feature labels that are clustered into the same category form the first-level clustering labels of all sites in the site set.

[0138] Step 505: If the number of currently accumulated cluster labels does not reach the quantity threshold, calculate the intra-class variance of each site category in the first-level cluster labels.

[0139] The quantity threshold refers to the upper limit of the site category.

[0140] Step 506: Use the second-level selected features to perform next-level clustering on the sites under the site category with the largest intra-class variance in the first-level clustering labels, and obtain second-level clustering labels for all sites under the site category.

[0141] The next level of clustering may refer to steps 501 to 504 and will not be described in detail.

[0142] Step 507: If the number of the currently accumulated cluster labels does not reach the quantity threshold, calculate the intra-class variance of each site category in the secondary cluster labels.

[0143] Step 508: Use the third-level selected features to perform next-level clustering on the sites under the site category with the largest intra-class variance in the second-level clustering labels, and obtain the third-level clustering labels of all sites under the site category.

[0144] The next level of clustering may refer to steps 501 to 504 and will not be described in detail.

[0145] Step 509, and so on, are performed until the number of tags of the currently accumulated cluster tags reaches the number threshold.

[0146] refer to Figure 6 As shown, in some other embodiments of the present specification, when the hierarchical selected features are continuous features, clustering sites using the selected features of each level in a hierarchical order from high to low may include the following steps:

[0147] Step 601: When the first-level selected feature is a continuous feature, hierarchical clustering of non-oil sales volume of the site is used to obtain a temporary class label of the site.

[0148] The first-level selected features are continuous features, which means that all the selected features at the first level are continuous features.

[0149] Step 602: Estimate the distribution of continuous feature values ​​of sites that are clustered into the same category to obtain multiple normal distributions.

[0150] In the embodiments of this specification, the distribution of continuous feature values ​​of sites clustered into the same category can be fitted with a normal distribution, so multiple normal distributions can be obtained. For example, if the sites are clustered into K categories, there are K normal distributions.

[0151] Step 603: Arrange the multiple normal distributions in order, and calculate the intersection of density functions of every two adjacent normal distributions.

[0152] For example, if there are K normal distributions, then K-1 intersection points can be obtained.

[0153] Step 604: Divide the first-level selected features into multiple intervals according to the intersection points, and cluster the sites in the same interval into one category, so as to obtain the first-level clustering labels of all sites in the site set.

[0154] Step 605: If the number of currently accumulated cluster labels does not reach the quantity threshold, calculate the intra-class variance of each site category in the first-level cluster labels.

[0155] Step 606: Use the second-level selected features to perform next-level clustering on the sites under the site category with the largest intra-class variance in the first-level clustering labels, and obtain the second-level clustering labels of all sites under the site category.

[0156] Step 607: If the number of currently accumulated cluster labels does not reach the quantity threshold, calculate the intra-class variance of each site category in the secondary cluster labels.

[0157] Step 608: Use the third-level selected features to perform next-level clustering on the sites under the site category with the largest intra-class variance in the second-level clustering labels, and obtain third-level clustering labels for all sites under the site category.

[0158] Step 609, and so on, are performed until the number of tags of the currently accumulated cluster tags reaches the quantity threshold.

[0159] like Figure 7 As shown, in an exemplary embodiment, based on Figure 4-Figure 6 The method shown can be Figure 7 The site collection in is divided into 7 site categories (in Figure 7 The second and fifth categories of sites are invalid categories, and the rest of the sites are valid categories).

[0160] refer to Figure 8 As shown, in some embodiments of the present specification, according to the non-oil product feature data and non-oil product inventory data of each site under each site category, the initial category structure of each site category is correspondingly determined, which may include performing the following operation steps for each site category:

[0161] Step 801: Determine the commodity number vector of all sites under the current site category, and take the maximum value in the commodity number vector as the maximum number of commodities for sale under the current site category.

[0162] You can first calculate the product number vector of all sites (a vector with only one column); for each site category, you can extract the data of all sites belonging to this category from the product number vector to form a product number vector for all sites under this site category, and then take the maximum value of the product number vector for all sites under this site category as the maximum number of products R sold under this site category.

[0163] Step 802: construct a commodity sales information matrix for this site category based on the non-oil inventory data of all sites under this site category.

[0164] In some embodiments of the present specification, the elements in the commodity sale information matrix of a site category can be expressed as:

[0165]

[0166] in, is the value of the jth product on sale at the ith site in the product sales information matrix of the kth site category.

[0167] Step 803: Based on the non-oil product feature data of all sites under this site category, a category matrix, a category brand matrix and a category price band matrix of this site category at a given category granularity are constructed respectively.

[0168] Category granularity refers to the coarseness or fineness of the classification of categories. For example, non-oil products can be divided into "cigarettes", "beverages" and other major categories according to major categories. If divided according to medium categories, "cigarettes" can be divided into "menthol cigarettes" and "non-menthol cigarettes"; if divided according to minor categories, "menthol cigarettes" can be divided into "imported menthol cigarettes" and "domestic menthol cigarettes", and "non-menthol cigarettes" can be divided into "imported non-menthol cigarettes" and "domestic non-menthol cigarettes".

[0169] In some embodiments of the present specification, an element of a site category in a category matrix of a given category granularity may be expressed as:

[0170]

[0171] in, is the value of the jth item belonging to the lth category in the category matrix of a given category granularity for a site category.

[0172] Category brand refers to the brand under a category; for example, taking the category of "cigarettes" as an example, the corresponding cigarette brands may include brand A cigarettes, brand B cigarettes, and brand C cigarettes, etc.

[0173] In some embodiments of the present specification, the elements of a site category in a category brand matrix of a given category granularity can be expressed as:

[0174]

[0175] in, It is the value of the j-th product belonging to the l-th category brand in the category brand matrix of a given category granularity for a site category.

[0176] Price Zone refers to the price range of a brand of goods in a category. For example, taking brand A cigarettes as an example, the price range of each box is 20-25 yuan.

[0177] In some embodiments of the present specification, the elements of a site category in a category price band matrix of a given category granularity can be expressed as:

[0178]

[0179] in, It is the value of the j-th product belonging to the l-th category price band in the category price band matrix of a given category granularity for a site category.

[0180] Step 804a, multiply the commodity sale information matrix by the category matrix to obtain a commodity number matrix of various categories sold at each site under this site category; the rows of the commodity number matrix of each category are sites and the columns are commodity categories; and based on the minimum value of each column in the commodity number matrix of each category, a commodity number matrix of the least sold category of this site category is formed as the first commodity number matrix.

[0181] For example, if the product sales information matrix is ​​Z and the category matrix is ​​D (1) , then the number matrix of commodities sold in each category at each site under a site category can be expressed as: H (1) =ZD (1) .

[0182] Step 804b, multiply the commodity sales information matrix with the category brand matrix to obtain a commodity number matrix of various category brands sold at each site under this site category; the behavior of the commodity number matrix of each category brand is the site and the column is the category brand; and according to the minimum value of each column in the commodity number matrix of each category brand, form a commodity number matrix of the least sold category brand of this site category as the second commodity number matrix.

[0183] For example, if the product sales information matrix is ​​Z and the category brand matrix is ​​D (2) , then the number matrix of commodities sold by each site under a site category can be expressed as: H (2) =ZD (2) .

[0184] Step 804c, multiply the commodity sale information matrix with the category price band matrix to obtain a commodity number matrix of each category price band sold at each site under this site category; the matrix of commodity number of each category price band is composed of sites and columns as category price bands; and based on the minimum value of each column in the matrix of commodity number of each category price band, a commodity number matrix of the least sold category price band of this site category is formed as the third commodity number matrix.

[0185] For example, if the product sales information matrix is ​​Z and the category price range matrix is ​​D (3) , then the number matrix of commodities sold in various price ranges at each site under a site category can be expressed as: H (3) =ZD (3) .

[0186] Step 805: The maximum number of sold commodities under the site category, the first commodity number matrix, the second commodity number matrix and the third commodity number matrix are taken together as the initial category structure of the site category.

[0187] In the embodiments of the present specification, the site category to which the site belongs, the profit (including theoretical profit and actual profit), sales volume (theoretical sales volume and actual sales volume) and inventory (theoretical inventory volume and actual inventory volume) of the goods sold by the site have a great influence on category optimization. Therefore, according to the initial category structure of each site category, as well as the non-oil feature data, non-oil sales volume data and category sales ratio of each site under each site category, a category optimization model can be created for each site category to optimize the category structure by classification.

[0188] In some embodiments of this specification, due to the uncertainty of demand, the estimation of theoretical sales volume (or theoretical demand) cannot simply give a certain value, but should give the probability of reaching the sales volume, that is, the demand distribution, so as to determine a reasonable theoretical sales volume, and accurately estimate the turnover rate and out-of-stock probability of the product. Specifically, the distribution fitting of historical sales volume data can be performed based on the mixed negative binomial distribution method, and then the 95% quantile of the fitted distribution is taken as the theoretical sales volume of the product.

[0189] In addition, it should be noted that if a site has never sold a certain product before and its historical sales data is 0, then the theoretical demand estimated based on the distribution fitting method is 0. However, this does not necessarily mean that the site will not make a profit by selling this product. Therefore, in order to solve the inaccuracy of historical sales data, it is necessary to use machine learning methods to further fill in the missing values ​​in the sales estimation matrix. For products and sites without historical sales data (the product has not been sold on the site, or it is a new product / site), according to the assumption, the demand estimation results of similar products and similar sites can be used to predict sales.

[0190] Specifically, for a given site i and product j, the following four cases can be considered:

[0191] (1) Product j has not been sold at site i, but has been sold at other sites. The site is taken as a sample, and the demand prediction results of each product at the site and the site characteristics are taken as sample characteristics, so that k1 similar sites of site i in the same category (such as gas station category) can be determined. If the number of neighbors in the same category is less than k1, the remaining similar sites are searched without category restriction, and the demand prediction results of product j at similar sites (if any) are averaged to obtain a demand prediction result; the product is taken as a sample, and the demand prediction results of the product at each site and the product characteristics are taken as sample characteristics, so that k2 similar products of the product in the same category (such as product category) can be determined. If the number of neighbors in the same category is less than k2, the remaining similar products are searched without category restriction, and the demand prediction results of similar products at site i (if any) are averaged to obtain a demand prediction result. The above two demand prediction results are averaged to obtain the demand prediction result of product j at site i.

[0192] (2) Site i is a brand new site, and the product has been sold at other sites. Based on the characteristics of the new site i, k1 similar sites are identified in the old sites with the same subcategory (such as convenience store category) as the new site. If there are less than k1, the remaining similar sites are identified in the old sites with the same major category as the new site. If there are still less than k1, the remaining neighbors are searched without category restriction. The demand forecast results of product j at similar sites (if any) are averaged to obtain the demand forecast results of product j at the new site i.

[0193] (3) Product j is a brand new product, and site i has sold other products. Based on the characteristics of new product j, k2 similar products are identified from old products with the same subcategory (such as product subcategory) as the new product. If there are less than k2, the remaining similar products are identified from old products with the same major category as the new product. If there are still less than k2, the remaining neighbors are searched without category restriction. The demand forecast results of similar products at site i (if any) are averaged to obtain the demand forecast result of new product j at site i.

[0194] (4) Site i is a brand new site, and product j is a brand new product. Similarly, according to the method of determining the nearest neighbor relationship in cases (2) and (3), k1 similar old sites of the new site i and k2 similar old products of the new product j can be determined, and the average of the demand prediction results (if any) of these k2 similar products at the k1 similar sites is used as the demand prediction result of the new product j at the new site i.

[0195] In some embodiments of this specification, the theoretical inventory can be estimated based on the theoretical sales volume. Let M be the expected sales volume per day, and the expected sales volume in different time periods can be calculated:

[0196] E(m1)=ML

[0197] E(m2)=M(TL)

[0198] E(m3)=ML

[0199] Among them, m1, m2, and m3 are the actual sales volume from the current category optimization decision to the current arrival of goods, the actual sales volume from the current arrival of goods to the next category optimization decision, and the actual sales volume from the next category optimization decision to the next arrival of goods, respectively. Fig. 9 shown.

[0200] Let the remaining inventory of the product at the time of this category optimization decision be D. If the replenishment strategy adopted is reasonable enough, the replenished inventory should be able to cover the demand for the product from the time the current replenishment decision is made to the next arrival. That is, the purchase quantity x should satisfy the inequality:

[0201] D+x≥A(L+T)

[0202] According to the lower bound calculation, the purchase quantity x should satisfy the following equation:

[0203] x=A(L+T)-D

[0204] Under equilibrium conditions, in order to maintain a stable inventory level, the replenishment quantity x should be equal to the expected sales volume within a decision cycle, that is:

[0205] x=E(m1+m2)=MT

[0206] So, we can get:

[0207] D=A(L+T)-x=AL+(AM)T

[0208] Therefore, the theoretical average inventory level in a replenishment cycle can be calculated as:

[0209]

[0210] Among them, A is the theoretical daily demand, M is the expected daily sales, L is the number of days for replenishment delay, and T is the replenishment cycle.

[0211] In some embodiments of the present specification, the theoretical profit can be obtained by calculating the gross profit rate of return based on the theoretical sales volume and the theoretical inventory volume.

[0212] In some embodiments of the present specification, the general expression of the category optimization model of each of the site categories can be expressed as:

[0213]

[0214] The corresponding constraints are:

[0215]

[0216] Among them, max is the maximum value objective function, x kj is whether the product j of the k-th site is sold, α is the hyperparameter that controls the weight of profit and sales, m is the number of products, n is the number of sites, K is, M j is the profit of commodity j, S ij is the demand for commodity j in the planning period at site i, T ik is whether site i belongs to the kth category site, p j is the proportion of sites where product j is sold at least in the system, For p j The product of n is rounded up, R k is the maximum number of commodities sold on the k-th category site, is whether product j belongs to the lth category, is the minimum number of commodities sold under the lth category in the kth category site, Is whether product j belongs to the lth category brand, is the minimum number of products sold under the brand of the lth category in the kth category site, Is whether product j belongs to the lth category price band, is the minimum number of commodities sold in the price band of the lth category in the kth category site, C ijis the purchase cost of product j in site i, G is the lower bound of gross profit rate of return, V ij is the forecast inventory for the planning period of item j in site i.

[0217] The above category optimization model is an integer linear programming model as a whole. However, since the sixth constraint is a nonlinear term, it can be converted into a linear term. In addition, due to the huge scale of variables in the above category optimization model, it can be solved by a heuristic algorithm based on a greedy adaptive random search strategy to improve the optimization efficiency. Under this solution algorithm, the category structure is first initialized, and then, under the constraints of the constraints, for the objective function (i.e., the category optimization model), the local optimal solution is selected in the feasible neighborhood each time.

[0218] Although the process flows described above include multiple operations that occur in a particular order, it should be clearly understood that the processes may include more or fewer operations, which may be performed sequentially or in parallel (eg, using parallel processors or a multi-threaded environment).

[0219] Corresponding to the above-mentioned gas station non-oil product category optimization method, the embodiment of this specification also provides a gas station non-oil product category optimization device, which can be configured on the above-mentioned service end, referring to Fig.10 As shown, in some embodiments of this specification, the non-oil product category optimization device of a gas station may include:

[0220] A site clustering module 101 is used to perform site clustering based on site feature data and non-oil product sales data of a site set to obtain multiple site categories;

[0221] A structure determination module 102, configured to determine an initial category structure for each site category according to the non-oil product characteristic data and non-oil product inventory data of each site under each site category;

[0222] The model creation module 103 is used to create a category optimization model for each site category according to the initial category structure of each site category, and the non-oil product feature data, non-oil product sales data and category sales ratio of each site under each site category;

[0223] The model solving module 104 is used to solve the category optimization model of each of the site categories according to the heuristic algorithm, and obtain the recommended category structure of each of the site categories.

[0224] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0225] It should be noted that in the embodiments of this specification, the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved are all information and data authorized by the user and fully authorized by all parties.

[0226] The embodiments of this specification also provide a computer device. Fig.11 As shown, in some embodiments of the present specification, the computer device 1102 may include one or more processors 1104, such as one or more central processing units (CPUs) or graphics processing units (GPUs), and each processing unit may implement one or more hardware threads. The computer device 1102 may also include any memory 1106, which is used to store any kind of information such as code, settings, data, etc. In a specific embodiment, the computer program on the memory 1106 and can be run on the processor 1104, when the computer program is run by the processor 1104, the instructions of the non-oil category optimization method of the gas station described in any of the above embodiments may be executed. Non-limitingly, for example, the memory 1106 may include any one or more combinations of the following: any type of RAM, any type of ROM, flash memory device, hard disk, optical disk, etc. More generally, any memory may use any technology to store information. Further, any memory may provide volatile or non-volatile retention of information. Further, any memory may represent a fixed or removable component of the computer device 1102. In one embodiment, when the processor 1104 executes the associated instructions stored in any memory or combination of memories, the computer device 1102 can perform any operation of the associated instructions. The computer device 1102 also includes one or more drive mechanisms 1108 for interacting with any memory, such as a hard disk drive mechanism, an optical disk drive mechanism, etc.

[0227] The computer device 1102 may also include an input / output interface 1110 (I / O) for receiving various inputs (via input devices 1112) and for providing various outputs (via output devices 1114). A specific output mechanism may include a presentation device 1116 and an associated graphical user interface 1118 (GUI). In other embodiments, the input / output interface 1110 (I / O), input device 1112, and output device 1114 may not be included, and the computer device 1102 may be used as a computer device in a network. The computer device 1102 may also include one or more network interfaces 1120 for exchanging data with other devices via one or more communication links 1122. One or more communication buses 1124 couple the components described above together.

[0228] The communication link 1122 may be implemented in any manner, for example, through a local area network, a wide area network (e.g., the Internet), a point-to-point connection, etc., or any combination thereof. The communication link 1122 may include any combination of hardwired links, wireless links, routers, gateway functions, name servers, etc. governed by any protocol or combination of protocols.

[0229] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), computer-readable storage media, and computer program products of some embodiments of the present specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processor to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processor generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0230] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processor to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0231] These computer program instructions can also be loaded onto a computer or other programmable data processor so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0232] In a typical configuration, a computer device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0233] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0234] Computer-readable media include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computer device. As defined in this specification, computer-readable media does not include temporary computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0235] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems or computer program products. Therefore, the embodiments of this specification may take the form of complete hardware embodiments, complete software embodiments or embodiments combining software and hardware. Moreover, the embodiments of this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0236] The present specification embodiments may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present specification embodiments may also be practiced in distributed computing environments where tasks are performed by remote processors connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0237] It should also be understood that in the embodiments of this specification, the term "and / or" is only a description of the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0238] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0239] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the embodiments of this specification. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification and the features of different embodiments or examples, without contradiction.

[0240] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for optimizing non-oil product categories at a gas station, characterized in that: include: Perform site clustering based on site feature data of site collection and non-oil product sales data to obtain multiple site categories; According to the non-oil product characteristic data and non-oil product inventory data of each site under each site category, an initial category structure of each site category is correspondingly determined; According to the initial category structure of each site category, as well as the non-oil product characteristic data, non-oil product sales data and category sales ratio of each site under each site category, a category optimization model for each site category is created accordingly; The category optimization model of each of the site categories is solved according to the heuristic algorithm, and the recommended category structure of each of the site categories is obtained accordingly.

2. The non-oil product category optimization method for a gas station according to claim 1, characterized in that: The site clustering based on the site characteristic data and non-oil product sales data of the site set includes: Performing correlation analysis on the full feature set formed by the site feature data based on the site set and the non-oil product sales data within a specified period to remove redundant features and form a selected feature set; determining a hierarchical level for the selected feature set; The sites are clustered using the selected features at each level in descending order.

3. The non-oil product category optimization method for a gas station as claimed in claim 2, characterized in that: Determining the hierarchical level of the selected feature set includes: Using the site features in the selected feature set as input variables, calling the mixed negative binomial distribution method to predict the non-oil product sales volume of the site, and determining the first-level selected features in the selected feature set according to the prediction results; Using the site features other than the first-level selected features in the selected feature set as input variables, calling the mixed negative binomial distribution method to predict the non-oil product sales volume of the site, and determining the second-level selected features in the selected feature set according to the prediction results; Using the site features other than the first-level selected features and the second-level selected features in the selected feature set as input variables, calling the mixed negative binomial distribution method to predict the non-oil product sales volume of the site, and determining the third-level selected features in the selected feature set according to the prediction results; The process is performed in this way until the selected features of a specified level in the selected feature set are obtained.

4. The non-oil product category optimization method for a gas station as claimed in claim 3, characterized in that: The site clustering is performed using the selected features of each level in a hierarchical order from high to low, including: When the first-level selected features are discrete features, hierarchical clustering is performed using the non-oil sales volume of the site to obtain the temporary class label of the site; Calculating a confusion matrix of discrete feature labels and the temporary class labels; the discrete feature labels are feature labels of the first level selected features; Performing hierarchical clustering on the discrete feature labels using the columns of the confusion matrix as features; Determining the primary clustering labels of all sites in the site set according to the hierarchical clustering results of the discrete feature labels; If the number of labels of the currently accumulated cluster labels does not reach the quantity threshold, calculating the intra-class variance of each site category in the first-level cluster labels; Using the second-level selected features, clustering the sites under the site category with the largest intra-class variance in the first-level clustering labels to the next level, and obtaining the second-level clustering labels of all sites under the site category; If the number of labels of the currently accumulated cluster labels does not reach the quantity threshold, calculating the intra-class variance of each site category in the secondary cluster labels; Using the third-level selected features, clustering the sites under the site category with the largest intra-class variance in the second-level clustering labels to the next level, and obtaining the third-level clustering labels of all sites under the site category; This process is repeated in this way until the number of currently accumulated cluster labels reaches the threshold.

5. The non-oil product category optimization method for a gas station as claimed in claim 3, characterized in that: The site clustering is performed using the selected features of each level in a hierarchical order from high to low, including: When the first-level selected feature is a continuous feature, the site non-oil sales hierarchical clustering is used to obtain the site temporary class label; The distribution of continuous feature values ​​of sites clustered into the same category was estimated, and multiple normal distributions were obtained; Arrange the multiple normal distributions in order, and calculate the intersection of density functions of every two adjacent normal distributions; Dividing the first-level selected features into multiple intervals according to the intersection points, and clustering the sites in the same interval into one category, to obtain the first-level clustering labels of all sites in the site set; If the number of labels of the currently accumulated cluster labels does not reach the quantity threshold, calculating the intra-class variance of each site category in the first-level cluster labels; Using the second-level selected features, clustering the sites under the site category with the largest intra-class variance in the first-level clustering labels to the next level, and obtaining the second-level clustering labels of all sites under the site category; If the number of labels of the currently accumulated cluster labels does not reach the quantity threshold, calculating the intra-class variance of each site category in the secondary cluster labels; Using the third-level selected features, clustering the sites under the site category with the largest intra-class variance in the second-level clustering labels to the next level, and obtaining the third-level clustering labels of all sites under the site category; This process is repeated in this way until the number of currently accumulated cluster labels reaches the threshold.

6. The non-oil product category optimization method for a gas station as claimed in claim 2, characterized in that: Determining the initial category structure of each site category according to the non-oil product feature data and non-oil product inventory data of each site under each site category includes: For each site category, do the following: Determine the commodity number vector of all sites under the site category, and take the maximum value in the commodity number vector as the maximum number of commodities for sale under the site category; Based on the non-oil inventory data of all sites under this site category, construct the commodity sales information matrix of this site category; Based on the non-oil product characteristic data of all sites under this site category, the category matrix, category brand matrix and category price band matrix of this site category at a given category granularity are constructed respectively; The commodity sale information matrix is ​​multiplied by the category matrix to obtain a commodity number matrix of each category sold at each site under the site category; the rows of the commodity number matrix of each category are sites and the columns are commodity categories; and according to the minimum value of each column in the commodity number matrix of each category, a commodity number matrix of the least sold category of the site category is formed as the first commodity number matrix; The commodity sales information matrix is ​​multiplied by the category brand matrix to obtain a commodity number matrix of various category brands sold at each site under the site category; the behavior of the commodity number matrix of each category brand is the site and the column is the category brand; and according to the minimum value of each column in the commodity number matrix of each category brand, a commodity number matrix of the least sold category brand of the site category is formed as the second commodity number matrix; The commodity sale information matrix is ​​multiplied by the category price band matrix to obtain a commodity number matrix of each category price band sold at each site under the site category; the matrix of commodity number of each category price band is composed of sites and columns as category price bands; and according to the minimum value of each column in the matrix of commodity number of each category price band, a commodity number matrix of the least sold category price band of the site category is formed as the third commodity number matrix; The maximum number of sold commodities under the site category, the first commodity number matrix, the second commodity number matrix and the third commodity number matrix are taken together as the initial category structure of the site category.

7. The non-oil product category optimization method for a gas station as claimed in claim 1, characterized in that: The general expression of the category optimization model for each site category is: The corresponding constraints are: Among them, max is the maximum value objective function, x kj is whether the product j of the k-th site is sold, α is the hyperparameter that controls the weight of profit and sales, m is the number of products, n is the number of sites, K is, M j is the profit of commodity j, S ij is the demand for commodity j in the planning period at site i, T ik is whether site i belongs to the kth category site, p j is the proportion of sites where product j is sold at least in the system, For p j The product of n is rounded up, R k is the maximum number of commodities sold on the k-th category site, is whether product j belongs to the lth category, is the minimum number of commodities sold under the lth category in the kth category site, Is whether product j belongs to the lth category brand, is the minimum number of products sold under the brand of the lth category in the kth category site, Is whether product j belongs to the lth category price band, is the minimum number of commodities sold in the price band of the lth category in the kth category site, C ij is the purchase cost of product j in site i, G is the lower bound of gross profit rate of return, V ij is the forecast inventory for the planning period of item j in site i.

8. A non-oil product category optimization device for a gas station, characterized in that: include: A site clustering module is used to cluster sites based on site feature data and non-oil product sales data of a site collection to obtain multiple site categories; A structure determination module, used for determining the initial category structure of each site category according to the non-oil product characteristic data and non-oil product inventory data of each site under each site category; A model creation module, for creating a category optimization model for each of the site categories according to the initial category structure of each of the site categories, and the non-oil product feature data, non-oil product sales data and category sales ratio of each site under each of the site categories; The model solving module is used to solve the category optimization model of each of the site categories according to the heuristic algorithm, and obtain the recommended category structure of each of the site categories.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory, characterized in that: When the computer program is executed by the processor, the computer program executes the instructions of the method according to any one of claims 1 to 7.

10. A computer storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor of a computer device, the computer program executes instructions of the method according to any one of claims 1 to 7.

11. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor of a computer device, the computer program executes instructions of the method according to any one of claims 1 to 7.

Images