Brand shop commercial site selection method based on sales prediction

Abstract

The invention relates to the technical field of intelligent business analysis, and particularly discloses a brand shop business site selection method based on sales prediction, which comprises the following steps: selecting shop sales and independent variable indexes influencing the shop sales according to different target brands, establishing an index system, and collecting LBS position data, customer portrait data, traffic data and POI data; the indexes are processed, the processing content comprises population radiation capacity calculation, traffic convenience calculation, trunk accessibility calculation and abnormal driving capacity calculation, correlation analysis is conducted on all the indexes and shop sales, indexes with high correlation are screened, a regression model is established according to the internal relation between independent variables and dependent variables, and a final model is determined; in combination with the user LBS position data and the customer portrait data, the accuracy and timeliness of the data can be ensured, and powerful technical support can be provided for intelligent shop operation site selection on the premise of protecting the privacy of the user.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent business analysis, and in particular relates to a method for selecting a brand store site based on sales forecast. Background Art

[0002] With the increasingly fierce business competition and the continuous changes in consumer demand, scientific and reasonable business site selection has become a key factor in the development of enterprises. The business site selection technologies currently on the market mainly include the following: Based on Geographic Information System (GIS): Combined with maps and geographic data, it helps to identify and analyze the characteristics of specific areas, such as population density, competitor distribution, infrastructure, etc., to assist in business site selection; Location-based social network (LBSN) data: Utilizes user check-in data on social networks to analyze users’ spatiotemporal behavior trajectories and preferences, thereby recommending suitable commercial store locations; Business district analysis method: involves investigation of the scope of effective business districts, investigation of business district customer groups, investigation of competing stores, etc., to provide data support for business site selection; Optimization algorithms: These include network optimization algorithms used to determine the best logistics layout, location selection to minimize operating costs or maximize revenue.

[0003] However, the above-mentioned commercial site selection technology is highly dependent on accurate and comprehensive data, such as LBSN data or GIS data, which may be limited by data quality and coverage, resulting in low accuracy and decision-making efficiency of commercial site selection. Therefore, we need to propose a brand store commercial site selection method based on sales forecast to solve the above-mentioned problems, so that it can use intelligent technology to improve the accuracy and decision-making efficiency of commercial site selection. Summary of the invention

[0004] The purpose of the present invention is to provide a brand store commercial site selection method based on sales forecast, which can use intelligent technology to improve the accuracy and decision-making efficiency of commercial site selection to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention adopts the following technical solutions: A method for selecting a brand store location based on sales forecast includes the following steps: S1. According to different target brands, select store sales and independent variable indicators that affect store sales, list indicators from three dimensions: store, business district, and geographical location, and build an indicator system; S2, collect LBS location data, customer profile data, traffic data and POI data; S3. Process the indicators, including the calculation of population radiation capacity, traffic convenience, trunk road accessibility and different traffic driving capacity; S4. Conduct correlation analysis between each indicator and store sales, and select indicators with strong correlation; S5. According to the intrinsic relationship between the independent variables and the dependent variables, a regression model is established to determine the final model; S6. Predict the store sales of candidate locations based on the regression model to guide the store location selection plan of the target brand.

[0006] Preferably, the LBS location data includes a person's anonymous ID, stop location, start time and end time; the customer portrait data includes marital status, age, gender, education level, children's age, income level, consumption level and occupation; the traffic data includes urban road network, parking lot data and bus and subway data; the POI data includes store name, store type, store area and store sales.

[0007] Preferably, the process of calculating the population radiation capacity is to extract the population of the target shopping mall respectively. ; Calculate the probability percentage of people visiting at each distance ,Right now ,in To extract the distance of the crowd; calculate the number of possible visitors for each distance crowd ,Right now , sum up the number of possible visitors at each distance, the summation formula is as follows: ,in, To extract the number of groups of people at different distances.

[0008] Preferably, the process of calculating the traffic convenience is as follows: A1. Define the radiation range: Set the radiation range of buses and subways to 1km; A2. Determine surrounding transportation facilities: Determine surrounding facilities of the target area based on the radiation range; A3. Calculate the distance from surrounding transportation facilities to the target shopping mall: normalize to score ; A4. Surrounding transportation facilities are divided into ; A5. Calculate the influence score of each surrounding transportation facility on the target shopping mall ,Right now = * ; A6. Calculate the total impact score of transportation facilities on the target shopping mall , and normalize: , then To standardize.

[0009] Preferably, the calculation of the main road accessibility is to use GIS to construct an OD cost matrix, calculate the average main road access time from each starting point to other end points in the city; then apply the interpolation method in GIS to obtain the average main road access time of the target location based on the average main road access time of the known location.

[0010] Preferably, the calculation process of the different-line driving capacity calculation is as follows: B1. Extract LBS location data within the specified range: Extract LBS location data from the target store to the visitor group 2 hours before and after entering the target store, and perform data cleaning; B2. Calculate the relevance of each level 3 business format of the target store: Count the number of customers who visit the target store, and summarize them by each level 3 business format to get the number of customers in each level 3 business format. , calculate the three-level format correlation of each target store , where j represents different three-level business formats; B3. Calculate the comprehensive relevance of the target stores: Summarize the relevance of each target store and calculate the comprehensive relevance value ; B4. Calculate the score driven by different industries in the target store buffer zone: Extract the brand details of each business format in the target store buffer zone, and standardize the distance between the target store and other stores to , calculate the target store's outlier driving score as , summarize the scores driven by abnormal transactions in the target store buffer .

[0011] Preferably, in step B4, when calculating the score driven by different behaviors in the buffer zone of the target store, if the target store is a store inside a shopping mall, the buffer zone is the same floor of the shopping mall, otherwise the buffer zone is determined based on the behavior trajectory of the customer 2 hours before and after visiting the target store.

[0012] Preferably, when screening the indicators, it is necessary to conduct a correlation analysis between each indicator and store sales, screen out indicators with strong correlation, and then conduct a multicollinearity test on the indicators to determine whether multicollinearity exists. If so, use principal component analysis to reduce the dimension.

[0013] Preferably, the regression model includes linear regression, forward / backward regression analysis, LASSO regression, ridge regression, and Bayesian regression.

[0014] The method for selecting a brand store location based on sales forecast proposed by the present invention has the following advantages compared with the prior art: 1. The present invention selects store sales and independent variable indicators that affect store sales according to different target brands, builds an indicator system, collects LBS location data, customer portrait data, traffic data and POI data; processes the indicators, and the processing content includes population radiation capacity calculation, traffic convenience calculation, main road accessibility calculation and abnormal driving capacity calculation, performs correlation analysis on each indicator and store sales, screens out indicators with strong correlation, establishes a regression model according to the intrinsic relationship between the independent variable and the dependent variable, determines the final model, predicts store sales at candidate locations according to the regression model, guides the store site selection plan of the target brand, and combines user LBS location data and customer portrait data to not only ensure the accuracy and timeliness of the data, but also provide strong technical support for intelligent store operation and site selection under the premise of protecting user privacy. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a flowchart of the present invention; Figure 2 A flowchart of the traffic convenience calculation of the present invention; Figure 3 It is a flowchart of the calculation process of the different-line driving capacity calculation of the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the accompanying drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. The specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0017] The present invention provides Figure 1-3 A method for selecting a brand store location based on sales forecast is shown, comprising the following steps: S1. According to different target brands, select store sales and independent variable indicators that affect store sales, list indicators from three dimensions: store, business district, and geographical location, and build an indicator system. The independent variable indicators include store, business district, business district customer portrait, geographical location, traffic convenience, average city arrival time, radiation population capacity, business atmosphere, and special travel belts; S2, collect LBS location data, customer profile data, traffic data and POI data; The LBS location data includes a person's anonymous ID, location, start time and end time; the customer portrait data includes marital status, age, gender, education level, children's age, income level, consumption level and occupation; the traffic data includes urban road network, parking lot data and bus and subway data; the POI data includes store name, store type, store area and store sales.

[0018] S3. Process the indicators, including the calculation of population radiation capacity, traffic convenience, trunk road accessibility and different traffic driving capacity; The process of calculating the population radiation capacity is to extract the population of the target shopping mall respectively. ; Calculate the probability percentage of people visiting at each distance ,Right now ,in To extract the distance of the crowd; calculate the number of possible visitors for each distance crowd ,Right now , sum up the number of possible visitors at each distance, the summation formula is as follows: ,in, To extract the number of groups of people at different distances; For example: extract the population within 1km, 1-2km, 2-3km…4-5km of the target mall respectively; calculate the probability percentage of people at each distance to visit: A=1 / distance, such as the probability percentage of people at 5km to visit A=1 / 5=20%; calculate the possible number of people visiting for each distance group B=A*population; sum up the possible number of people visiting for each distance group.

[0019] The process of calculating the traffic convenience is as follows: A1. Define the radiation range: Set the radiation range of buses and subways to 1km; A2. Determine surrounding transportation facilities: Determine surrounding facilities of the target area based on the radiation range; A3. Calculate the distance from surrounding transportation facilities to the target shopping mall: normalize to score ,The higher the score, the closer the distance and the greater the impact; A4. Surrounding transportation facilities are divided into ,For bus / subway stations, the quality score decreases gradually according to the number of lines; A5. Calculate the influence score of each surrounding transportation facility on the target shopping mall ,Right now = * ; A6. Calculate the total impact score of transportation facilities on the target shopping mall , and normalize: , then To standardize; For example: extract the names, longitudes and latitudes of all bus / subway station names in the target city, and the line information they belong to; obtain the bus / subway station information within 1km of the target shopping mall, and count the number of lines at each station. The bus station information needs to be deduplicated because there are round-trip routes; assign a value of 100 to the station with the most lines among all stations, and use the range indexation method to calculate the scores of other stations; add up the scores of all stations within 1km of the target shopping mall as the bus / subway convenience value.

[0020] The calculation of arterial road accessibility is to construct an OD cost matrix using GIS, calculate the average arterial road access time from each starting point to other end points in the city; then apply the interpolation method in GIS to obtain the average arterial road access time of the target location based on the average arterial road access time of the known location. The interpolation method in GIS refers to the inverse distance weighted method, which assumes that each sampling point has a certain local influence ability, and its influence weakens with the increase of distance. Points with closer distances have greater weights, which have a greater impact on the interpolation results; for example: if you need to predict the average arterial road access time in a certain area, you only need to use the points with known traffic time in the area as sample input, and by calculating their distances from the target point and the corresponding weights, you can estimate the traffic time of the target point.

[0021] The calculation process of the different-line driving capacity calculation is as follows: B1. Extract LBS location data within the specified range: Extract LBS location data from the target store to the visitor group 2 hours before and after entering the target store, and perform data cleaning; B2. Calculate the relevance of each level 3 business format of the target store: Count the number of customers who visit the target store, and summarize them by each level 3 business format to get the number of customers in each level 3 business format. , calculate the three-level format correlation of each target store , where j represents different three-level business formats; B3. Calculate the comprehensive relevance of the target stores: Summarize the relevance of each target store and calculate the comprehensive relevance value ; B4. Calculate the score driven by different industries in the target store buffer zone: Extract the brand details of each business format in the target store buffer zone, and standardize the distance between the target store and other stores to , calculate the target store's outlier driving score as , summarize the scores driven by abnormal transactions in the target store buffer ; For example, extract the LBS location data of the reference store to the visitor group, clean and obtain the details of all consumption scenarios visited within a time period. Count the number of customers who visited the relevant stores, and summarize them by three-level business types, calculate the three-level business type relevance of each target store, summarize the relevance of each target store, calculate the comprehensive relevance value, use this comprehensive relevance value as a measurement standard, and again judge the outlier driving ability of the mall where each target store is located.

[0022] When calculating the score driven by different behaviors within the buffer zone of the target store, if the target store is a store inside a shopping mall, the buffer zone is the same floor of the shopping mall; otherwise, the buffer zone is determined based on the customer's behavioral trajectory 2 hours before and after visiting the target store.

[0023] S4. Conduct correlation analysis between each indicator and store sales, and select indicators with strong correlation; When screening the indicators, it is necessary to conduct a correlation analysis between each indicator and store sales, screen out indicators with strong correlation, and then conduct a multicollinearity test on the indicators to determine whether there is multicollinearity. If so, use principal component analysis to reduce the dimension.

[0024] S5. According to the intrinsic relationship between the independent variables and the dependent variables, a regression model is established to determine the final model; The regression models include linear regression, forward / backward regression analysis, LASSO regression, ridge regression, and Bayesian regression.

[0025] Among them, the linear regression method assumes that there is a linear relationship between the dependent variable and the independent variable. The goal of linear regression is to find the best fitting line (or hyperplane) to minimize the prediction error; The forward / backward regression analysis optimizes the model by gradually adding or removing variables, thus improving the accuracy and reliability of the regression analysis. LASSO regression and ridge regression limit model complexity and prevent overfitting by adding penalty terms to the loss function; Bayesian regression estimates the posterior distribution of model parameters by combining prior knowledge and observed data; S6. Predict the store sales of candidate locations based on the regression model to guide the store location selection plan of the target brand. The specific operations are as follows: input the prepared indicator data into the trained regression model, run the regression model, obtain the predicted sales of each candidate location, and then analyze the accuracy and reliability of the prediction results, check whether there is overfitting or underfitting, compare the predicted sales of different locations, evaluate their conformity with the actual market situation, consider the overall brand development strategy, determine which locations’ predicted results are consistent with the brand positioning, evaluate the cost budget of each candidate location, including rent, decoration, labor costs, etc., compare with the predicted sales, determine the return on investment, consider current market trends, such as changes in consumer preferences, competitor layout, future development potential, etc., select the best location, conduct a comprehensive analysis of sales forecasts, cost budgets, brand strategies and market trends, weigh various factors, and select the best store location for opening based on the results of the comprehensive analysis.

[0026] In summary, by selecting store sales and independent variable indicators that affect store sales according to different target brands, an indicator system is established, and LBS location data, customer portrait data, traffic data and POI data are collected; the indicators are processed, including the calculation of population radiation capacity, traffic convenience, main road accessibility and different-traffic driving capacity, and the correlation between each indicator and store sales is analyzed, and indicators with strong correlation are screened. According to the intrinsic relationship between the independent variable and the dependent variable, a regression model is established to determine the final model, and the store sales of the candidate locations are predicted according to the regression model to guide the store site selection plan of the target brand. Combining user LBS location data and customer portrait data can not only ensure the accuracy and timeliness of the data, but also provide strong technical support for intelligent store operation and site selection under the premise of protecting user privacy.

[0027] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for selecting a brand store location based on sales forecast, characterized by: The steps include: S1. According to different target brands, select store sales and independent variable indicators that affect store sales, list indicators from three dimensions: store, business district, and geographical location, and build an indicator system; S2, collect LBS location data, customer profile data, traffic data and POI data; S3. Process the indicators, including the calculation of population radiation capacity, traffic convenience, trunk road accessibility and different traffic driving capacity; S4. Conduct correlation analysis between each indicator and store sales, and select indicators with strong correlation; S5. According to the intrinsic relationship between the independent variables and the dependent variables, a regression model is established to determine the final model; S6. Predict the store sales of candidate locations based on the regression model to guide the store location selection plan of the target brand.

2. The method for selecting a brand store location based on sales forecast according to claim 1, characterized in that: The LBS location data includes a person's anonymous ID, location, start time and end time; the customer portrait data includes marital status, age, gender, education level, children's age, income level, consumption level and occupation; the traffic data includes urban road network, parking lot data and bus and subway data; the POI data includes store name, store type, store area and store sales.

3. The method for selecting a brand store location based on sales forecast according to claim 2, characterized in that: The process of calculating the population radiation capacity is to extract the population of the target shopping mall respectively. ; Calculate the probability percentage of people visiting at each distance ,Right now ,in To extract the distance of the crowd; calculate the number of possible visitors for each distance crowd ,Right now , sum up the number of possible visitors at each distance, the summation formula is as follows: ,in, To extract the number of groups of people at different distances.

4. The method for selecting a brand store location based on sales forecast according to claim 3, characterized in that: The process of calculating the traffic convenience is as follows: A1. Define the radiation range: Set the radiation range of buses and subways to 1km; A2. Determine surrounding transportation facilities: Determine surrounding facilities of the target area based on the radiation range; A3. Calculate the distance from surrounding transportation facilities to the target shopping mall: normalize to score ; A4. Surrounding transportation facilities are divided into ; A5. Calculate the influence score of each surrounding transportation facility on the target shopping mall ,Right now = * ; A6. Calculate the total impact score of transportation facilities on the target shopping mall , and normalize: , then To standardize.

5. The method for selecting a brand store location based on sales forecast according to claim 4, characterized in that: The trunk road accessibility calculation is to use GIS to construct an OD cost matrix, calculate the average trunk road access time from each starting point to other destinations in the city; then use the interpolation method in GIS to obtain the average trunk road access time of the target location based on the average trunk road access time of the known location.

6. A method for selecting a brand store location based on sales forecast according to claim 5, characterized in that: The calculation process of the different-line driving capacity calculation is as follows: B1. Extract LBS location data within the specified range: Extract LBS location data from the target store to the visitor group 2 hours before and after entering the target store, and perform data cleaning; B2. Calculate the relevance of each level 3 business format of the target store: Count the number of customers who visit the target store, and summarize them by each level 3 business format to get the number of customers in each level 3 business format. , calculate the three-level format correlation of each target store , where j represents different three-level business formats; B3. Calculate the comprehensive relevance of the target stores: Summarize the relevance of each target store and calculate the comprehensive relevance value ; B4. Calculate the score driven by different industries in the target store buffer zone: Extract the brand details of each business format in the target store buffer zone, and standardize the distance between the target store and other stores to , calculate the target store's outlier driving score as , summarize the scores driven by abnormal transactions in the target store buffer .

7. A method for selecting a brand store location based on sales forecast according to claim 6, characterized in that: In step B4, when calculating the score driven by different behaviors in the buffer zone of the target store, if the target store is a store inside the mall, the buffer zone is the same floor of the mall, otherwise the buffer zone is determined based on the behavior trajectory of the customer 2 hours before and after visiting the target store.

8. The method for selecting a brand store location based on sales forecast according to claim 7, characterized in that: When screening the indicators, it is necessary to conduct a correlation analysis between each indicator and store sales, screen out indicators with strong correlation, and then conduct a multicollinearity test on the indicators to determine whether there is multicollinearity. If so, use principal component analysis to reduce the dimension.

9. The method for selecting a brand store location based on sales forecast according to claim 8, characterized in that: The regression models include linear regression, forward / backward regression analysis, LASSO regression, ridge regression, and Bayesian regression.

Images