Spatial clustering analysis-based dot layout method and apparatus, and electronic device

Abstract

The invention provides a website layout method and device based on spatial clustering analysis and electronic equipment, and relates to the field of website layout, and the method comprises the steps: determining the types of all target products according to historical sales volume data, carrying out the grid processing of a target hot spot region, determining each sub-grid region, and carrying out the grid processing of the target hot spot region; calculating a variance value between the grid clustering center and a standard grid clustering center, and analyzing and processing the sub-grid regions by using kernel density to obtain an aggregation region; and adjusting the sales volume of each target product by using the variance value and the aggregation region to obtain the adjusted sales volume of each target product, determining the order quantity of the target products according to the target POI, determining the fishing net region with the maximum order quantity as a target network point, and determining the product configuration quantity of each target product. According to the method, store layout site selection, product selection and product quantity configuration can be realized, spatial distribution of store site selection factors is evaluated by adopting spatial clustering, the order quantity of a fishing net area is calculated, and a more reasonable network layout is given.

Description

Technical Field

[0001] The present invention relates to the field of network point layout, and in particular to a network point layout method, device and electronic equipment based on spatial clustering analysis. Background Art

[0002] The location theory originated from classical economics. It is a combination theory of spatial selection and economic activities within the space. The location theory is widely used in various aspects such as urban land use, factory site selection, commercial service network establishment, urban system and regional planning. It has two connotations: one is the spatial selection of economic behavior; the other is the organic combination of economic activities within the space. The former can be called layout location theory, and the latter can be called business location theory. There are certain differences between these two location theories in terms of research procedures and methods. The layout location theory is that the location subject is known, and based on the inherent characteristics of the location subject itself, it analyzes the possible space suitable for the location subject, and then selects the best location from it. The business location theory is just the opposite. The large location space is known, and the optimal combination and spatial form of the location subject are studied based on the geographical characteristics, economic and social conditions of the space.

[0003] The classic site selection method needs to consider population factors, business environment, traffic factors and competition factors. Among the population factors, the main ones are population mobility and population density, which can serve as an important reference for the marketing of fast-moving consumer goods; in the business environment, the main considerations are the number and distance of commercial POIs. If the scale can be obtained, the weight is assigned; in the traffic factor, the recommended stores are set up on both sides of the road. In areas with the same suitability, the stores are set up on the side of the road with a higher level; in the competition factor, in a limited area, too much and too fierce competition will make retailers pay a huge price. The defects or difficulties of the classic site selection method come from the implementation level. For example, it is difficult to obtain data statistically. The flow of people in the population factor can be counted by squatting, but when there are many alternative locations, the cost will be very high. As for the consumption habits, average income and other data in the population factor, it is impossible to obtain them directly by statistics; for example, the weight is difficult to quantify objectively. For the business environment and traffic factors, the expert scoring method can be used to quantify, but the scoring results are also subject to a relatively large subjective influence.

[0004] However, there is currently no technical solution that can solve the above technical problems, and there is no network point layout method, device and electronic equipment based on spatial clustering analysis. Summary of the invention

[0005] The present invention provides a network point layout method, device and electronic equipment based on spatial clustering analysis, which constructs sales space clustering through store sales data, predicts the potential sales of each network point in each region, and determines the network point layout and the product configuration quantity of the target product in the laid out network point to meet the actual needs of users.

[0006] In a first aspect, the present invention provides a network point layout method based on spatial clustering analysis, comprising:

[0007] From the target hotspot area, determine the types of all target products according to the historical sales data, wherein the target hotspot area is determined according to each heat map corresponding to all candidate products in the preset area, and each heat map is determined according to all historical sales data of each candidate product in different stores;

[0008] Using K-means clustering algorithm to process the historical sales data of all target products in the target hotspot area to obtain the standard grid cluster center, gridding the target hotspot area, determining each sub-grid area, for each sub-grid area, using K-means clustering algorithm to process the historical sales data of all target products in the sub-grid area to obtain the sales of each target product in the grid cluster center, calculating the variance value between the grid cluster center and the standard grid cluster center, using kernel density analysis to process the sub-grid area to obtain the clustering area of ​​each target product;

[0009] The sales volume of each target product is adjusted by using the variance value and the clustering area to obtain the adjusted sales volume of each target product in the subgrid area, the subgrid area is divided into all fishing net areas, the order quantity of each target product is determined according to the target POI of each fishing net area, each fishing net area in all subgrid areas is traversed, and the fishing net area with the largest order quantity is determined as the target network point;

[0010] The product configuration quantity of each target product in the target outlet is determined according to the adjusted sales volume of each target product corresponding to the target outlet and the other store sales volume of each target product corresponding to other stores in the subgrid area where the target outlet is located.

[0011] According to the network layout method based on spatial cluster analysis provided by the present invention, the target product is a fast-moving consumer product, and the method of determining the types of all target products based on historical sales data includes:

[0012] From the target hotspot area, identify all stores with increasing sales in the historical months;

[0013] Identify the product categories of the top three products in each of the sales growth stores across all sales growth stores;

[0014] The cumulative quantity of each product category in all stores with sales growth is counted, and the three product categories with the largest cumulative quantity are determined as the categories of all target products.

[0015] According to the network layout method based on spatial clustering analysis provided by the present invention, the historical sales data of all target products in the target hotspot area are processed using the K-means clustering algorithm to obtain the standard grid cluster center, including:

[0016] Building a data set based on the historical sales data of all target products in the target hotspot area, wherein the historical sales data includes the sales volume of each store in different historical months;

[0017] A preset number of cluster centers are randomly determined from the data set, the distance from each data point to any cluster center is calculated, and each data point is assigned to the cluster where the nearest cluster center is located, and the cluster centers are updated again until a preset number of iterations is reached, and all updated cluster centers are determined as the standard grid cluster centers.

[0018] According to the network point layout method based on spatial clustering analysis provided by the present invention, the gridding process of the target hot spot area to determine each sub-grid area includes:

[0019] When it is determined that the store density of the target hot spot area is less than the first preset density, the target hot spot area is divided according to the first preset side length to determine each sub-grid area;

[0020] When it is determined that the store density of the target hot spot area is greater than or equal to the first preset density and less than the second preset density, the target hot spot area is divided according to the second preset side length to determine each sub-grid area;

[0021] When it is determined that the store density of the target hot spot area is greater than or equal to the second preset density, dividing the target hot spot area according to the third preset side length to determine each sub-grid area;

[0022] Among them, the first preset side length is greater than the second preset side length, and the second preset side length is greater than the third preset side length.

[0023] According to the network point layout method based on spatial cluster analysis provided by the present invention, the calculation of the variance value between the grid cluster center and the standard grid cluster center includes:

[0024]

[0025] Among them, M(i) is the variance value, xn(i), yn(i), zn(i) are the spatial coordinates of any grid cluster center, and Xn, Yn, Zn are the spatial coordinates of the standard grid cluster center.

[0026] According to the network layout method based on spatial clustering analysis provided by the present invention, the use of the variance value and the clustering area to adjust the sales volume of each target product to obtain the adjusted sales volume of each target product in the sub-grid area includes:

[0027] determining a plurality of levels in the sub-grid area according to the aggregation area;

[0028] For each level, the sales volume of each target product is adjusted by adjusting the variance value to obtain the adjusted sales volume of each target product of each level in the sub-grid area.

[0029] According to the network point layout method based on spatial cluster analysis provided by the present invention, the order quantity of each target product is determined according to the target POI of each fishing net area, including:

[0030] T=P ij1 *X1+P ij2 *X2+P ij3 *X3

[0031] Among them, T is the order quantity of the target product, P ij1 is the preset factor affected by commercial POI, P ij2 is the preset factor affected by the service POI, P ij3 is the preset factor affected by traffic POI, X1 is the distance to commercial POI, X2 is the distance to service POI, and X3 is the distance to traffic POI.

[0032] According to the network layout method based on spatial clustering analysis provided by the present invention, the product configuration quantity of each target product in the target network is determined according to the adjusted sales volume of each target product corresponding to the target network and the other store sales volume of each target product corresponding to other stores in the subgrid area where the target network is located, including:

[0033] For each target product, determine the difference between the adjusted sales volume and the sales volume of other stores, and determine the product configuration quantity of the target product;

[0034] All target products are traversed to determine the product configuration quantity of each target product in the target network point.

[0035] In a second aspect, a network point layout device based on spatial clustering analysis is provided, comprising:

[0036] a first determination unit, the first determination unit being used to determine the types of all target products from a target hotspot area according to historical sales data, wherein the target hotspot area is determined in a preset area according to each heat map corresponding to all candidate products, and each heat map is determined according to all historical sales data of each candidate product in different stores;

[0037] A processing unit, the processing unit is used to process the historical sales data of all target products in the target hot spot area by using a K-means clustering algorithm to obtain a standard grid cluster center, grid-process the target hot spot area, determine each sub-grid area, and for each sub-grid area, process the historical sales data of all target products in the sub-grid area by using a K-means clustering algorithm to obtain the sales volume of each target product in the grid cluster center, calculate the variance value between the grid cluster center and the standard grid cluster center, and process the sub-grid area by using a kernel density analysis to obtain a clustering area for each target product;

[0038] A division unit, the division unit is used to adjust the sales volume of each target product by using the variance value and the clustering area, obtain the adjusted sales volume of each target product in the sub-grid area, divide the sub-grid area into all fishing net areas, determine the order quantity of each target product according to the target POI of each fishing net area, traverse each fishing net area in all sub-grid areas, and determine the fishing net area with the largest order quantity as the target network point;

[0039] The second determination unit is used to determine the product configuration quantity of each target product in the target outlet according to the adjusted sales volume of each target product corresponding to the target outlet and the other store sales volume of each target product corresponding to other stores in the sub-grid area where the target outlet is located.

[0040] In a third aspect, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the network layout method based on spatial clustering analysis when executing the program.

[0041] The present invention combines the first law of geography and location theory, that is, the influence of factors such as human flow, population characteristics, business district and traffic conditions on store sales is fed back to the spatial location distribution, and then the sales space clustering is constructed only through a large number of store sales data in different locations and different time periods, and based on this, a spatial prediction model of store sales is established in combination with the distribution of various POIs in space to provide calculation support for the layout optimization of marketing outlets. The present invention aims at current fast-moving consumer goods, realizes store layout site selection, product selection and product quantity configuration, uses historical sales data to provide sufficient data sources for network layout site selection, and adopts sales space clustering to evaluate the spatial distribution of store site selection factors, breaking through the previous need to collect statistical data such as human flow and consumption habits, and the problem that it is impossible to effectively quantify and evaluate the impact of business districts and traffic;

[0042] The present invention calculates the cluster centers of different types of commodities with large differences in consumption distribution, deduces the spatial distribution of consumer groups, and then combines the division of spatial grids with the calculation of grid variance, uses the distribution of consumption characteristics to determine the quantity demand gap of target products in different areas, and calculates the order quantity of a smaller area within the sub-grid area, namely the fishing net area, in combination with the distribution of various POIs, to provide a more reasonable network layout. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0044] Figure 1 It is a schematic diagram of the flow of the network point layout method based on spatial clustering analysis provided by the present invention;

[0045] Figure 2 is a thermal map provided by the present invention;

[0046] Figure 3 It is a flow chart of counting the cumulative quantity of each product category in all stores with sales growth provided by the present invention;

[0047] Figure 4 It is a schematic diagram of the cluster center provided by the present invention;

[0048] Figure 5 It is a schematic diagram of total sales volume and variance provided by the present invention;

[0049] Figure 6 It is a schematic diagram of the relationship between the grid average sales volume and the mean square error provided by the present invention;

[0050] Figure 7 is a schematic diagram of adjusting the sales volume of each target product by adjusting the variance value provided by the present invention;

[0051] Figure 8 is a schematic diagram of determining a target network point provided by the present invention;

[0052] Fig. 9 It is a structural schematic diagram of a network point layout device based on spatial clustering analysis provided by the present invention;

[0053] Fig.10 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0055] Figure 1 : is a schematic diagram of a flow chart of a network point layout method based on spatial clustering analysis provided by the present invention, wherein the network point layout method based on spatial clustering analysis comprises:

[0056] Step 101: Determine the types of all target products from the target hotspot area according to the historical sales data. The target hotspot area is determined in a preset area according to each heat map corresponding to all candidate products. Each heat map is determined according to all historical sales data of each candidate product in different stores.

[0057] Step 102: Process the historical sales data of all target products in the target hotspot area using the K-means clustering algorithm to obtain the standard grid cluster center, grid the target hotspot area, determine each sub-grid area, and for each sub-grid area, process the historical sales data of all target products in the sub-grid area using the K-means clustering algorithm to obtain the sales volume of each target product in the grid cluster center, calculate the variance value between the grid cluster center and the standard grid cluster center, and process the sub-grid area using kernel density analysis to obtain the clustering area of ​​each target product;

[0058] Step 103: adjusting the sales volume of each target product using the variance value and the clustering area to obtain the adjusted sales volume of each target product in the subgrid area, dividing the subgrid area into all fishing net areas, determining the order quantity of each target product according to the target POI of each fishing net area, traversing each fishing net area in all subgrid areas, and determining the fishing net area with the largest order quantity as the target network point;

[0059] Step 104: determine the product configuration quantity of each target product in the target outlet according to the adjusted sales volume of each target product corresponding to the target outlet and the other store sales volume of each target product corresponding to other stores in the subgrid area where the target outlet is located.

[0060] In step 101, firstly, the historical sales data of all candidate products are collected from the preset area, including sales records of different stores and different time periods. Figure 2 is the thermal diagram provided by the present invention, such as Figure 2As shown, first select appropriate historical sales data, unify the data format and measurement unit, and then count the sales amount / volume of each type of goods at the store and month granularity, and then form a heat map of each type of goods.

[0061] For each candidate product, a heat map is drawn on the map based on its historical sales data to show the sales popularity of the product in different stores. According to the distribution of the heat map, areas with higher sales popularity are identified as target hot spots. Within the target hot spots, products with growth potential are screened out as target products based on factors such as the growth trend of historical sales data, and the types of all target products are determined. The sales distribution of candidate products is intuitively displayed through the drawing of the heat map, which helps to quickly identify hot spots. It is also possible to directly select the types of goods with significantly different sales hot spots by visual inspection. The screened target products have clear positioning characteristics and growth potential, which provides strong support for subsequent spatial clustering analysis and network layout.

[0062] Optionally, the target product is a fast-moving consumer product, and determining the types of all target products according to historical sales data includes:

[0063] From the target hotspot area, identify all stores with increasing sales in the historical months;

[0064] Identify the product categories of the top three products in each of the sales growth stores across all sales growth stores;

[0065] The cumulative quantity of each product category in all stores with sales growth is counted, and the three product categories with the largest cumulative quantity are determined as the categories of all target products.

[0066] Figure 3 : is a flow chart of counting the cumulative quantity of each product category in all stores with sales growth provided by the present invention, such as Figure 3 As shown, the present invention first collects historical sales data of all stores in the target hot spot area in the past period of time (such as the past year or a specific month), and based on the collected data, screens out stores that have shown a sales growth trend in historical months, namely, sales growth stores. By comparing the sales data of each month, stores with continuous sales growth or a significant growth trend can be identified. For each sales growth store, the three products with the highest sales are determined based on the historical sales data, and the screened products are classified according to the product categories they belong to, such as food, beverages, daily necessities, etc. For each product category, the cumulative number of products of this category in all sales growth stores is counted, and based on the statistical results of the cumulative number, the three product categories with the largest cumulative number are selected as the types of all target products.

[0067] In an optional embodiment, the number of stores with increasing sales and decreasing sales of each product within a certain period of time is counted separately to exclude the impact of long holidays on sales fluctuations, such as January and October, wherein the selection criteria for increasing stores are:

[0068]

[0069] Find the stores whose M is greater than 0, that is, the growing stores, where M is the average monthly sales fluctuation compared to January, n is the number of statistical months in the year (usually 12), Si is the sales of the i-th month (excluding January and October), and S1 is the sales of January.

[0070] Among them, the screening basis for reducing stores is:

[0071]

[0072] Find the stores whose M is less than 0, that is, reduce the number of stores, where M is the average monthly sales fluctuation compared to October, n is the number of statistical months in the year (usually 12), Si is the sales of the i-th month (excluding January and October), and S10 is the sales of October.

[0073] The present invention compares the spatial distribution of stores with growth and stores with decline of various commodities, selects commodity types with significant differences, and through combining the screening of stores with sales growth, the determination of hot-selling product categories and the statistics of cumulative quantities, it can accurately identify the types of fast-moving consumer goods with the greatest consumer demand and the strongest growth potential in the target hot spot area. It not only has the characteristics of being data-driven, but also can flexibly adapt to market changes and consumer needs in different regions.

[0074] In step 102, the historical sales data of all target products in the target hotspot area are clustered using the K-means clustering algorithm to obtain the standard grid cluster center; then, the target hotspot area is gridded to determine each sub-grid area, and in each sub-grid area, the historical sales data of the target product is processed again using the K-means clustering algorithm to obtain the sales of the target product in the grid cluster center, and the variance value between the grid cluster center of each sub-grid area and the standard grid cluster center is calculated to evaluate the degree of difference between the sub-grid area and the overall sales trend, and each sub-grid area is processed using kernel density analysis to obtain the clustering area of ​​the target product, that is, the sales hotspot or high-density area.

[0075] Optionally, the K-means clustering algorithm is used to process the historical sales data of all target products in the target hotspot area to obtain a standard grid cluster center, including:

[0076] Building a data set based on the historical sales data of all target products in the target hotspot area, wherein the historical sales data includes the sales volume of each store in different historical months;

[0077] A preset number of cluster centers are randomly determined from the data set, the distance from each data point to any cluster center is calculated, and each data point is assigned to the cluster where the nearest cluster center is located, and the cluster centers are updated again until a preset number of iterations is reached, and all updated cluster centers are determined as the standard grid cluster centers.

[0078] Figure 4 is a schematic diagram of the cluster center provided by the present invention, such as Figure 4 As shown in the figure, the historical sales data of all target products are collected from the target hotspot area. These data should include the sales volume of each store in different historical months to ensure the comprehensiveness and accuracy of the data. The collected data are preprocessed, such as removing outliers and filling missing values, to ensure the quality of the data and the accuracy of the analysis; the preprocessed data are organized into a data set suitable for processing by the K-means clustering algorithm.

[0079] Optionally, a preset number of data points are randomly selected from the data set as initial cluster centers. These initial cluster centers will serve as the starting point of the algorithm iteration. For each data point in the data set, its distance to all cluster centers is calculated, and the data point is assigned to the cluster where the nearest cluster center is located. Based on the data points in each cluster, the cluster center of each cluster is recalculated. The above steps are repeated until a preset number of iterations is reached or the change in the cluster center is less than a certain threshold, to ensure that the algorithm can converge to a stable solution. All updated cluster centers are determined as standard grid cluster centers for subsequent gridding processing and clustering analysis of sub-grid areas.

[0080] The present invention realizes K-means cluster analysis of the historical sales data of all target products in the target hotspot area by constructing a data set, randomly determining cluster centers, and iteratively updating the steps, and can accurately identify the natural clustering structure of sales data, providing strong support for subsequent grid processing and cluster analysis of sub-grid areas. At the same time, by selecting the preset number of iterations and the number of cluster centers, the accuracy and efficiency of the algorithm can be flexibly adjusted to adapt to data sets of different sizes and complexities.

[0081] In another optional embodiment, the k-means clustering algorithm is first used to perform clustering calculations according to the product category-sales volume of each store to obtain multiple clusters of two-dimensional clustering, and then the stores are classified into stores with the best sales volume of different product categories by using the silhouette coefficient. The calculation process of K-means spatial clustering is not repeated here. The silhouette coefficient s of sample i in the evaluation cluster isi , refer to the following formula:

[0082]

[0083] Among them, a i is the intra-cluster dissimilarity, that is, the average distance from sample i to each sample in the cluster. i is the out-of-cluster dissimilarity, that is, the average distance from sample i to each sample in other clusters. After 100 k-means clustering iterations, the cluster center is obtained as shown in Figure 4.

[0084] Optionally, determine each sub-grid area, and in each sub-grid area, use the K-means clustering algorithm again to process the historical sales data of the target product to obtain the sales volume of the target product in the grid cluster center:

[0085] Cluster Center Product TypeA Product Type B Product Type C Cluster Center 1 148 114 148 Cluster Center 2 62 34 78 Cluster Center 3 17 12 20

[0086] Optionally, the gridding process of the target hotspot area to determine each sub-grid area includes:

[0087] When it is determined that the store density of the target hot spot area is less than the first preset density, the target hot spot area is divided according to the first preset side length to determine each sub-grid area;

[0088] When it is determined that the store density of the target hot spot area is greater than or equal to the first preset density and less than the second preset density, the target hot spot area is divided according to the second preset side length to determine each sub-grid area;

[0089] When it is determined that the store density of the target hot spot area is greater than or equal to the second preset density, dividing the target hot spot area according to the third preset side length to determine each sub-grid area;

[0090] Among them, the first preset side length is greater than the second preset side length, and the second preset side length is greater than the third preset side length.

[0091] Optionally, the present invention first collects the location information of all stores in the target hot spot area to calculate the store density. According to the location information of the stores, the store density in the target hot spot area is calculated. The store density can be measured by the ratio of the number of stores to the area of ​​the region. If the store density in the target hot spot area is less than the first preset density, it means that the stores in the region are sparsely distributed. At this time, a larger first preset side length is used to divide the target hot spot area to determine each sub-grid area. The larger side length can reduce the number of sub-grids and improve processing efficiency; if the store density in the target hot spot area is greater than or equal to the first preset density and less than the second preset density, it means that the stores in the region are moderately distributed. At this time, a smaller second preset side length is used to divide the target hot spot area to determine each sub-grid area. The moderate side length can ensure that each sub-grid contains an appropriate number of stores, which is convenient for subsequent analysis; if the store density in the target hot spot area is greater than or equal to the second preset density, it means that the stores in the region are densely distributed. At this time, a smaller third preset side length is used to divide the target hot spot area to determine each sub-grid area. The smaller side length can divide the area more finely and improve the accuracy of the analysis.

[0092] In an optional embodiment, the grid is divided according to the density of stores, and then the variance of each grid is counted against the cluster center of the grid with stores. Generally, a large grid with a side length of 300 meters can be used for city-level stores, a medium grid with a side length of 150 meters can be used for street-level stores, and a small grid with a side length of 70 meters can be used for community-level stores. Then, the sales volume and cluster center of various types of goods in the stores within the grid are counted. By dynamically adjusting the side length of the grid division according to the store density, the flexibility and accuracy of the grid processing can be ensured. In low-density areas, larger side lengths can reduce the amount of calculation; in high-density areas, smaller side lengths can improve the precision of the analysis, thus achieving flexible grid processing of target hotspot areas.

[0093] Figure 5 is a schematic diagram of total sales volume and variance provided by the present invention, wherein the calculation of the variance value between the grid cluster center and the standard grid cluster center includes:

[0094]

[0095] Among them, M(i) is the variance value, xn(i), yn(i), n(i) are the grid cluster center coordinates of each month, Xn, Yn, Zn are the standard grid cluster center coordinates, calculate the difference evaluation factor (monthly variance) M between the cluster center of each month and the standard cluster center, set the cluster center coordinates of each month to xn(i), yn(i), zn(i) to evaluate the difference between the sub-grid area and the overall sales trend. The calculation of the variance value provides a basis for evaluating the sales difference of the sub-grid area, which is helpful for subsequent sales adjustment and network layout optimization. According to the statistics of sales and variance, it can be seen that as the variance decreases, the total sales increase. After a simple linear regression, the monthly variance can be used to guide the overall product ratio.

[0096] Furthermore, combined with the above formula, the variance M of the i-th grid to the standard cluster center is calculated by referring to the monthly variance statistics. i , the calculation formula is as follows:

[0097]

[0098] Among them, n is the total number of grids, x, y, and z are the cluster centers of the three products respectively, and x i ,y i ,z i are the cluster centers of the i-th grid respectively.

[0099] In step 103, the sales volume of the target product in each sub-grid area is adjusted according to the variance value and the clustering area. The sales volume adjustment can be performed based on factors such as sales volume trends and market demand. Each sub-grid area is further divided into multiple fishing net areas, each of which has similar sales volume characteristics and trends. According to the target POI (such as commercial facilities, population density, etc.) of each fishing net area and the adjusted sales volume, the order quantity of each target product is calculated, and all fishing net areas are traversed to select the fishing net area with the largest order volume as the target point.

[0100] Optionally, the adjusting the sales volume of each target product by using the variance value and the clustering area to obtain the adjusted sales volume of each target product in the subgrid area includes:

[0101] determining a plurality of levels in the sub-grid area according to the aggregation area;

[0102] For each level, the sales volume of each target product is adjusted by adjusting the variance value to obtain the adjusted sales volume of each target product of each level in the sub-grid area.

[0103] Figure 6 This is a schematic diagram of the relationship between the average grid sales volume and the mean square error provided by the present invention. Figure 7This is a schematic diagram of adjusting the sales volume of each target product by adjusting the variance value provided by the present invention. Through the relationship between grid sales volume and variance, sales volume is inversely proportional to variance within a specific interval. The variance M is changed to M_new = (1 / M). The larger the M_new, the smaller the variance and the closer to the standard ratio center. A kernel density analysis is performed on it, and the areas where M_new presents spatial aggregation are extracted and divided into multiple levels, and the sales ratio can be increased around them.

[0104] First, it is necessary to identify areas with significant sales aggregation characteristics in the sub-grid area, namely, aggregation areas, which can be determined by analyzing sales data, customer flow, geographical location and other factors. According to the characteristics of the aggregation area, such as size, sales intensity, customer distribution, etc., the sub-grid area is divided into multiple levels. The higher the level, the higher the sales aggregation degree in the area, and the sales volume of the target product may also be greater. For each level, the variance value of the sales volume of each target product in the level is calculated. The variance value reflects the degree of discreteness of the sales distribution. The larger the variance, the more uneven the sales distribution. According to the size of the variance value, the sales volume of each target product in each level is adjusted. Specifically, if the variance value is large, it means that the sales distribution is uneven, and it may be necessary to balance the sales volume through promotion, increase inventory and other methods; if the variance value is small, it means that the sales distribution is relatively uniform, and it may not be necessary to make a large-scale sales adjustment. After the above adjustment, the adjusted sales volume of each target product in each level of the sub-grid area is obtained, and the adjusted sales volume more accurately reflects the actual sales volume of each target product in the area. The present invention adjusts sales volume according to variance values ​​and hierarchical structures, and can more flexibly cope with sales volume differences in different regions, thereby improving the accuracy of prediction and the effectiveness of strategies.

[0105] Strategy 1: For each layer, adjust the M_new of stores with M_new lower than 8.1, 5.6 or 2.9 to 8.1 / 5.6 / 2.9, that is, reduce the gap with the standard cluster center. Then use the previous formula to calculate the theoretical sales volume for matching. The following table shows the sales improvement of this strategy;

[0106] Strategy 2: For each layer, further increase its M_new by 10-15% on the basis of 2.1 or 3.6 on 5.9 to get closer to the center of the ratio, and then query the theoretical sales volume and ratio according to the formula:

[0107] Store level Actual average sales volume Strategy 1 Average Sales Strategy 2 Average Sales First floor 604 628 655 Second floor 415 438 457 Third floor 308 325 364 Improvement rate 0 5.10％ 13.50％

[0108] Optionally, determining the order quantity of each target product according to the target POI of each fishing net area includes:

[0109] T=P ij1 *X1+P ij2 *X2+Pij3 *X3

[0110] Among them, T is the order quantity of the target product, P ij1 is the preset factor affected by commercial POI, P ij2 is the preset factor affected by the service POI, P ij3 is the preset factor affected by traffic POI, X1 is the distance to commercial POI, X2 is the distance to service POI, and X3 is the distance to traffic POI.

[0111] Different from the general site selection that models the entire area, in order to improve the accuracy of the optimized site selection, the present invention further divides the grid into different fishing net (BUFFER) areas according to regional segmentation (BUFFER), and establishes unique models for different areas; at the same time, the selection of BUFFER also ensures the maximum difference between regions and has the best effect. There are many categories of POI, with commercial, service, and transportation POIs as the key considerations; in this way, the commercial, service, and transportation attractions in each fishing net M are PM1, PM2, and PM3 respectively, and the user's purchase choice is inversely proportional to the distance from the POI. According to the law of gravity, to be precise, it is inversely proportional to the square of the distance, set to D. Then, for any product store numbered i in a circular surface, the factor affected by the surrounding POI is P ij .

[0112] According to the above formula:

[0113] serial number Commercial POI distance Traffic POI distance Service POI distance Order Quantity Sales 0 112.1 0 109.2 3330 550107 1 123.7 0 108.2 2945 522719 4 118.5 0 118.3 2615 411444 6 129.9 0 105.6 2165 372491

[0114] Calculate the distance from the center of each fishing net to the surrounding POIs, and plug it into the formula to calculate the theoretical sales volume, and get the fishing net with the maximum sales volume. That is, it is considered that it is more reasonable to set up stores inside these fishing nets, and their product ordering volume can be roughly predicted.

[0115] In step 104, other store sales data of each target product of other stores in the subgrid area where the target outlet is located is collected. For each target product, the difference between its adjusted sales at the target outlet and the sales at other stores is calculated. The product configuration quantity of each target product at the target outlet is determined based on factors such as the size of the difference and market demand. In an optional embodiment, the product configuration quantity of each target product in the target outlet is determined based on the adjusted sales of each target product corresponding to the target outlet and the other store sales of each target product corresponding to other stores in the subgrid area where the target outlet is located, including:

[0116] For each target product, determine the difference between the adjusted sales volume and the sales volume of other stores, and determine the product configuration quantity of the target product;

[0117] All target products are traversed to determine the product configuration quantity of each target product in the target network point.

[0118] Figure 8 It is a schematic diagram of determining target outlets provided by the present invention. The order volumes corresponding to the two fishing net numbers in areas 34 and 44 in the figure are the highest, which are around areas 34 and 44 in the figure; it is reflected on the map as follows: (the fishing net in area 34 in the figure already includes a store with an order volume of 3330, indicating that the demonstration result is feasible) the first four fishing nets are: areas 34, 44, 36, and 66 in the figure, and the optimization measure is to open three marketing outlets in areas 36, 44, and 66 in the figure. Referring to the map, it is found that area 66 in the figure is not near the road network and is eliminated. The conclusion is to open stores in areas 36 and 44 in the figure.

[0119] The network layout method based on spatial clustering analysis of the present invention realizes in-depth analysis of target product sales data, accurate identification of spatial clustering, reasonable optimization of sales adjustment and scientific planning of network layout through a series of specific implementation plans and technical means, which jointly improves the accuracy and effectiveness of network layout. According to location theory, factors affecting store site selection include surrounding fixed population, floating population, population structure, consumption habits, income situation, etc., which are in turn affected by transportation, residential areas, etc. These data are either difficult to objectively quantify or the cost of statistical acquisition is very high. The present invention combines the relevant theories of classical economics and geographic information science, constructs sales space clustering through store marketing data to predict the potential sales of each network in each region, and calculates the optimization of network layout.

[0120] The present invention is based on the combination of geographic information science and classical economics, and adopts the sales spatial clustering method to construct an analysis model, which can effectively avoid the above two adverse effects and directly use more objective and easier to obtain historical sales data of outlets as the basis for analysis.

[0121] As another optional embodiment of the present invention, the present invention is mainly divided into three steps for optimizing the network layout: first, calculate the spatial cluster center, select multiple commodity types with the largest possible sales spatial distribution difference according to the sales data of different commodities in each network point, and calculate the cluster center of each spatial cluster under different commodity ratios according to the spatial relationship. Due to the different population composition, consumption habits and income levels, the feedback to the store sales will lead to significant differences in sales amount / sales volume of different types of commodities. Therefore, the cluster center of the spatial cluster under different ratios objectively reflects the actual distribution of specific consumer groups in space; then, divide the spatial grid, and divide the spatial grid according to the distribution density of the sales network points for the sales area. Then calculate the variance of the average sales amount of the grid and the average sales amount of the cluster center, so as to infer the gap between the current commodity ratio and the ideal ratio in this grid area. By adjusting the distribution of the commodity ratio in the grid, the sales potential of the local area can be further excavated. Since the distribution of the specific consumer group is only similar within a certain range, the division of the spatial grid can better quantify this consumption feature. The cluster center is the most ideal distribution area of ​​consumption characteristics, and the variance from the cluster center is the gap from this ideal distribution. By narrowing this gap (not a completely linear relationship), we can further fit the distribution of consumption characteristics and transform potential consumer demand; finally, the site selection optimization within the grid is aimed at the store location within the grid, which needs to be further optimized in combination with the specific distribution of various POIs (commercial, transportation, etc.). First, further divide the grid into smaller fishing nets. According to the public POI data, calculate its anti-gravity impact factor on each fishing net center, including the drainage factor of commercial and transportation POIs and the competition factor of similar stores. Then, based on the multivariate regression of the sales sample, estimate the fishing net with the highest expected sales, and combine field investigations to determine the store location.

[0122] Based on the above technical solution, the present invention solves the problem of difficulty in selecting the location of fast-moving consumer goods stores. In view of the large amount of population data collection and statistical requirements required by the classic location selection algorithm, as well as the problem of being unable to objectively quantify the business district and traffic impact, the present invention only needs to collect and organize the historical sales data of each store in the outlet (classified and counted by commodity type), and then combine the public POI data to carry out the location selection calculation, without spending a lot of time and investing high costs to collect third-party data; exclude the influence of expert subjective factors in the location selection analysis and calculation as much as possible, and obtain alternative addresses with smaller alternative areas and higher accuracy through reasonable and objective calculations; through the constructed spatial clustering and the calculation of the grid variance (with the variance of the cluster center), it can further guide the reasonable distribution of goods in the grid area to tap the sales potential. In addition, the calculation results of the present invention can also be combined for sales forecasting, abnormal perception warning and other management optimization aspects.

[0123] Fig. 9 It is a structural schematic diagram of the network layout device based on spatial clustering analysis provided by the present invention, and the network layout device based on spatial clustering analysis includes a first determination unit 1, and the first determination unit is used to determine the types of all target products from the target hot spot area according to the historical sales data. The target hot spot area is determined in a preset area according to each heat map corresponding to all candidate products, and each heat map is determined according to all historical sales data of each candidate product in different stores. The working principle of the first determination unit 1 can refer to the aforementioned step 101 and will not be repeated here.

[0124] The network layout device based on spatial clustering analysis also includes a processing unit 2, which is used to use a K-means clustering algorithm to process the historical sales data of all target products in the target hot spot area to obtain a standard grid cluster center, grid-process the target hot spot area, determine each sub-grid area, and for each sub-grid area, use a K-means clustering algorithm to process the historical sales data of all target products in the sub-grid area to obtain the sales of each target product in the grid cluster center, calculate the variance value between the grid cluster center and the standard grid cluster center, and use kernel density analysis to process the sub-grid area to obtain the clustering area of ​​each target product. The working principle of the processing unit 2 can refer to the aforementioned step 102 and will not be repeated here.

[0125] The network point layout device based on spatial clustering analysis also includes a division unit 3, which is used to adjust the sales volume of each target product using the variance value and the clustering area to obtain the adjusted sales volume of each target product in the sub-grid area, divide the sub-grid area into all fishing net areas, determine the order quantity of each target product according to the target POI of each fishing net area, traverse each fishing net area in all sub-grid areas, and determine the fishing net area with the largest order quantity as the target network point. The working principle of the division unit 3 can refer to the aforementioned step 103 and will not be repeated here.

[0126] The network layout device based on spatial clustering analysis also includes a second determination unit 4, which is used to determine the product configuration quantity of each target product in the target network according to the adjusted sales volume of each target product corresponding to the target network and the other store sales volume of each target product corresponding to other stores in the sub-grid area where the target network is located. The working principle of the second determination unit 4 can refer to the aforementioned step 104 and will not be repeated here.

[0127] The present invention combines the first law of geography and location theory, that is, the influence of factors such as human flow, population characteristics, business district and traffic conditions on store sales is fed back to the spatial location distribution, and then the sales space clustering is constructed only through a large number of store sales data in different locations and different time periods, and based on this, a spatial prediction model of store sales is established in combination with the distribution of various POIs in space to provide calculation support for the layout optimization of marketing outlets. The present invention aims at current fast-moving consumer goods, realizes store layout site selection, product selection and product quantity configuration, uses historical sales data to provide sufficient data sources for network layout site selection, and adopts sales space clustering to evaluate the spatial distribution of store site selection factors, breaking through the previous need to collect statistical data such as human flow and consumption habits, and the problem that it is impossible to effectively quantify and evaluate the impact of business districts and traffic;

[0128] The present invention calculates the cluster centers of different types of commodities with large differences in consumption distribution, deduces the spatial distribution of consumer groups, and then combines the division of spatial grids with the calculation of grid variance, uses the distribution of consumption characteristics to determine the quantity demand gap of target products in different areas, and calculates the order quantity of a smaller area within the sub-grid area, namely the fishing net area, in combination with the distribution of various POIs, to provide a more reasonable network layout.

[0129] Fig.10 Schematic diagram of the structure of the electronic device provided by the present invention. Fig.10As shown, the electronic device may include: a processor 110, a communication interface 120, a memory 130 and a communication bus 140, wherein the processor 110, the communication interface 120 and the memory 130 communicate with each other through the communication bus 140. The processor 110 may call the logic instructions in the memory 130 to execute the network layout method based on spatial clustering analysis, the method comprising: determining the types of all target products from the target hotspot area according to the historical sales data, the target hotspot area is determined in a preset area according to each heat map corresponding to all candidate products, each heat map is determined according to all historical sales data of each candidate product in different stores; using the K-means clustering algorithm to process the historical sales data of all target products in the target hotspot area to obtain a standard grid cluster center, gridding the target hotspot area to determine each sub-grid area, for each sub-grid area, using the K-means clustering algorithm to process the historical sales data of all target products in the sub-grid area to obtain each target product in the grid cluster center. sales volume, calculate the variance value between the grid cluster center and the standard grid cluster center, process the sub-grid area using kernel density analysis to obtain the clustering area of ​​each target product; adjust the sales volume of each target product using the variance value and the clustering area to obtain the adjusted sales volume of each target product in the sub-grid area, divide the sub-grid area into all fishing net areas, determine the order quantity of each target product according to the target POI of each fishing net area, traverse each fishing net area in all sub-grid areas, and determine the fishing net area with the largest order quantity as the target network point; determine the product configuration quantity of each target product in the target network point according to the adjusted sales volume of each target product corresponding to the target network point and the other store sales volume of each target product corresponding to other stores in the sub-grid area where the target network point is located.

[0130] In addition, the logic instructions in the above-mentioned memory 130 can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0131] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute a network layout method based on spatial clustering analysis provided by the above methods, the method including: determining the types of all target products from the target hot spot area according to historical sales data, the target hot spot area is determined in a preset area according to each heat map corresponding to all candidate products, and each heat map is determined according to all historical sales data of each candidate product in different stores; using a K-means clustering algorithm to process the historical sales data of all target products in the target hot spot area to obtain a standard grid cluster center, gridding the target hot spot area, determining each sub-grid area, and for each sub-grid area, using a K-means clustering algorithm to process the sub-grid area. The historical sales data of all target products in the grid area are used to obtain the sales volume of each target product in the grid cluster center, the variance value between the grid cluster center and the standard grid cluster center is calculated, and the sub-grid area is processed by kernel density analysis to obtain the clustering area of ​​each target product; the sales volume of each target product is adjusted by using the variance value and the clustering area to obtain the adjusted sales volume of each target product in the sub-grid area, and the sub-grid area is divided into all fishing net areas, and the order quantity of each target product is determined according to the target POI of each fishing net area, and each fishing net area in all sub-grid areas is traversed to determine the fishing net area with the largest order quantity as the target network point; the product configuration quantity of each target product in the target network point is determined according to the adjusted sales volume of each target product corresponding to the target network point and the other store sales volume of each target product corresponding to other stores in the sub-grid area where the target network point is located.

[0132] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the network layout method based on spatial clustering analysis provided by the above-mentioned methods, the method comprising: determining the types of all target products from the target hot spot area according to historical sales data, the target hot spot area is determined in a preset area according to each heat map corresponding to all candidate products, and each heat map is determined according to all historical sales data of each candidate product in different stores; using a K-means clustering algorithm to process the historical sales data of all target products in the target hot spot area to obtain a standard grid clustering center, gridding the target hot spot area, determining each sub-grid area, and for each sub-grid area, using a K-means clustering algorithm to process the historical sales data of all target products in the sub-grid area The sales volume data of each target product in the grid cluster center is obtained, the variance value between the grid cluster center and the standard grid cluster center is calculated, and the sub-grid area is processed by kernel density analysis to obtain the clustering area of ​​each target product; the sales volume of each target product is adjusted by using the variance value and the clustering area to obtain the adjusted sales volume of each target product in the sub-grid area, and the sub-grid area is divided into all fishing net areas, and the order quantity of each target product is determined according to the target POI of each fishing net area, and each fishing net area in all sub-grid areas is traversed to determine the fishing net area with the largest order quantity as the target network point; the product configuration quantity of each target product in the target network point is determined according to the adjusted sales volume of each target product corresponding to the target network point and the other store sales volume of each target product corresponding to other stores in the sub-grid area where the target network point is located.

[0133] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0134] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0135] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A network point layout method based on spatial clustering analysis, characterized in that: include: From the target hotspot area, determine the types of all target products according to the historical sales data, wherein the target hotspot area is determined according to each heat map corresponding to all candidate products in the preset area, and each heat map is determined according to all historical sales data of each candidate product in different stores; Using K-means clustering algorithm to process the historical sales data of all target products in the target hotspot area to obtain the standard grid cluster center, gridding the target hotspot area, determining each sub-grid area, for each sub-grid area, using K-means clustering algorithm to process the historical sales data of all target products in the sub-grid area to obtain the sales of each target product in the grid cluster center, calculating the variance value between the grid cluster center and the standard grid cluster center, using kernel density analysis to process the sub-grid area to obtain the clustering area of ​​each target product; The sales volume of each target product is adjusted by using the variance value and the clustering area to obtain the adjusted sales volume of each target product in the subgrid area, the subgrid area is divided into all fishing net areas, the order quantity of each target product is determined according to the target POI of each fishing net area, each fishing net area in all subgrid areas is traversed, and the fishing net area with the largest order quantity is determined as the target network point; The product configuration quantity of each target product in the target outlet is determined according to the adjusted sales volume of each target product corresponding to the target outlet and the other store sales volume of each target product corresponding to other stores in the subgrid area where the target outlet is located.

2. The network point layout method based on spatial clustering analysis according to claim 1, characterized in that: The target product is a fast-moving consumer product, and the types of all target products are determined based on historical sales data, including: From the target hotspot area, identify all stores with increasing sales in the historical months; Identify the product categories of the top three products in each of the sales growth stores across all sales growth stores; The cumulative quantity of each product category in all stores with sales growth is counted, and the three product categories with the largest cumulative quantity are determined as the categories of all target products.

3. The network point layout method based on spatial clustering analysis according to claim 1, characterized in that: The K-means clustering algorithm is used to process the historical sales data of all target products in the target hotspot area to obtain the standard grid cluster center, including: Building a data set based on the historical sales data of all target products in the target hotspot area, wherein the historical sales data includes the sales volume of each store in different historical months; A preset number of cluster centers are randomly determined from the data set, the distance from each data point to any cluster center is calculated, and each data point is assigned to the cluster where the nearest cluster center is located, and the cluster centers are updated again until a preset number of iterations is reached, and all updated cluster centers are determined as the standard grid cluster centers.

4. The network point layout method based on spatial clustering analysis according to claim 1, characterized in that: The gridding process of the target hotspot area to determine each sub-grid area includes: When it is determined that the store density of the target hot spot area is less than the first preset density, the target hot spot area is divided according to the first preset side length to determine each sub-grid area; When it is determined that the store density of the target hot spot area is greater than or equal to the first preset density and less than the second preset density, the target hot spot area is divided according to the second preset side length to determine each sub-grid area; When it is determined that the store density of the target hot spot area is greater than or equal to the second preset density, dividing the target hot spot area according to the third preset side length to determine each sub-grid area; Among them, the first preset side length is greater than the second preset side length, and the second preset side length is greater than the third preset side length.

5. The network point layout method based on spatial clustering analysis according to claim 1, characterized in that: The calculating the variance value between the grid cluster center and the standard grid cluster center comprises: Among them, M(i) is the variance value, xn(i), yn(i), zn(i) are the spatial coordinates of any grid cluster center, and Xn, Yn, Zn are the spatial coordinates of the standard grid cluster center.

6. The network point layout method based on spatial clustering analysis according to claim 1, characterized in that: The step of adjusting the sales volume of each target product by using the variance value and the clustering area to obtain the adjusted sales volume of each target product in the subgrid area includes: determining a plurality of levels in the sub-grid area according to the aggregation area; For each level, the sales volume of each target product is adjusted by adjusting the variance value to obtain the adjusted sales volume of each target product of each level in the sub-grid area.

7. The network point layout method based on spatial clustering analysis according to claim 1, characterized in that: Determining the order quantity of each target product according to the target POI of each fishing net area includes: T=P ij1 *X1+P ij2 *X2+P ij3 *X3 Among them, T is the order quantity of the target product, P ij1 is the preset factor affected by commercial POI, P ij2 is the preset factor affected by the service POI, P ij3 is the preset factor affected by traffic POI, X1 is the distance to commercial POI, X2 is the distance to service POI, and X3 is the distance to traffic POI.

8. The network point layout method based on spatial clustering analysis according to claim 1, characterized in that: Determining the product configuration quantity of each target product in the target outlet according to the adjusted sales volume of each target product corresponding to the target outlet and other store sales volume of each target product corresponding to other stores in the subgrid area where the target outlet is located includes: For each target product, determine the difference between the adjusted sales volume and the sales volume of other stores, and determine the product configuration quantity of the target product; All target products are traversed to determine the product configuration quantity of each target product in the target network point.

9. A network point layout device based on spatial clustering analysis, characterized in that: include: a first determination unit, the first determination unit being used to determine the types of all target products from a target hotspot area according to historical sales data, wherein the target hotspot area is determined in a preset area according to each heat map corresponding to all candidate products, and each heat map is determined according to all historical sales data of each candidate product in different stores; A processing unit, the processing unit is used to process the historical sales data of all target products in the target hot spot area by using a K-means clustering algorithm to obtain a standard grid cluster center, grid-process the target hot spot area, determine each sub-grid area, and for each sub-grid area, process the historical sales data of all target products in the sub-grid area by using a K-means clustering algorithm to obtain the sales volume of each target product in the grid cluster center, calculate the variance value between the grid cluster center and the standard grid cluster center, and process the sub-grid area by using a kernel density analysis to obtain a clustering area for each target product; A division unit, the division unit is used to adjust the sales volume of each target product by using the variance value and the clustering area, obtain the adjusted sales volume of each target product in the sub-grid area, divide the sub-grid area into all fishing net areas, determine the order quantity of each target product according to the target POI of each fishing net area, traverse each fishing net area in all sub-grid areas, and determine the fishing net area with the largest order quantity as the target network point; The second determination unit is used to determine the product configuration quantity of each target product in the target outlet according to the adjusted sales volume of each target product corresponding to the target outlet and the other store sales volume of each target product corresponding to other stores in the sub-grid area where the target outlet is located.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the network point layout method based on spatial clustering analysis as described in any one of claims 1 to 8 is implemented.

Images