Method for evaluating store by combining relevant indexes of sorrounding environment of store
Patent Information
- Application Number
- MYPI2025005869
- Authority / Receiving Office
- MY · MY
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-03-27
- Filing Date
- 2024-03-26
- Publication Date
- 2024-09-27
AI Technical Summary
Existing store assessment methods rely on personnel on-site inspections and expert assessments, resulting in high information collection costs, low efficiency, and subjective results, making them unsuitable for large-scale regional assessments.
Through statistical modeling and linear regression models, the geographical information and natural features around the store are quantified, and geographical environment theme scores with business value are generated. Combined with the theme intensity and aggregation model, the store operating conditions and location requirements are evaluated.
It achieves an efficient and objective assessment of the store's surrounding environment, reduces personnel costs, eliminates the influence of subjective factors, improves assessment efficiency and accuracy, and provides a quantitative business reference basis.
Smart Images

Figure PI2025005869_DRAWING_001
Abstract
Description
A method for evaluating stores by combining relevant indexes of the store's surrounding environment Technical Field
[0001] The present invention relates to the technical field of store business evaluation, and in particular to a method for evaluating a store by combining relevant indexes of the store's surrounding environment. Background Art
[0002] When businesses conduct store site selection, store profiling, and offline store comparison testing, they typically need an objective understanding of the geographic environment, business environment, and population surrounding the evaluation point to facilitate their assessment. Store spatiotemporal indicators are a series of business-information-enhancing large-scale geographic environment thematic scores and related indices measuring travel convenience and comfort, derived through statistical modeling of surrounding geographic information and weather data. These indices perfectly meet the needs of businesses, helping them gain an objective and clear understanding of the store's surroundings and improving the efficiency of their assessment work.
[0003] Most existing assessment work involves collecting geographic information through field visits by personnel, and then using expert assessment voting to obtain regional geographic environment scores. The on-site information collection by personnel is extremely time-consuming in terms of personnel costs and time costs, and cannot guarantee that the information is collected completely. It cannot be used as an assessment method for large areas. In addition, the geographic environment scores obtained by expert scoring are too subjective, and the assessment results are seriously affected by personal factors.
[0004] Summary of the Invention
[0005] The present invention provides a method for evaluating stores by combining relevant indexes of the store's surrounding environment, which can effectively solve the problems proposed in the above background technology that on-site information collection by personnel is extremely time-consuming in terms of personnel cost and time cost, and cannot guarantee complete information collection, and cannot be used as an evaluation method for large-scale areas. In addition, the geographical environment score obtained by expert scoring is too subjective, and the evaluation results are seriously affected by personal factors.
[0006] To achieve the above objectives, the present invention provides the following technical solutions: a method for evaluating stores by combining relevant indices of the store's surrounding environment, which primarily quantifies a geographical environment theme score with business value and strong explanatory power from geographic information. This score, combined with geographic information and relevant natural features, is used to evaluate the store's operating conditions or ability to meet operating needs, providing a reference basis for store business scenarios.
[0007] The assessment steps include:
[0008] S1. Collect geographic information data around offline stores;
[0009] S2. Conduct statistical modeling of entities surrounding the store;
[0010] S3. Reasonably extract the subject feature index SSTI;
[0011] S4, linear regression model evaluation system;
[0012] S5. Indicator SSTI is applied to real business scenarios.
[0013] According to the above technical solution, in S1, offline store surrounding geographic information data is collected mainly by using different data sources to obtain geographic information around the target store, and the geographic information mainly includes the store's surrounding environment information and related natural features;
[0014] After obtaining geographic information, the facilities near the target store are divided into eight categories: transportation, residence, shopping, food, travel, life, work and entertainment. At the same time, for each theme, the number of theme facilities near the target store and the distance information to the target store are obtained.
[0015] According to the above technical solution, in S2, statistical modeling of the store surrounding entities is mainly carried out by establishing a topic intensity model and a topic aggregation model, and combining the results of the two models to generate a large spatiotemporal index under the topic;
[0016] This paper mainly uses the data of a certain brand as an example and combines it with a linear regression model to verify the impact of large-scale spatiotemporal indicators on customer flow. Once the verification is passed, the large-scale spatiotemporal indicators can be applied to the brand's real business scenarios to help improve the brand's evaluation efficiency of the store's surrounding environment, and ultimately serve as an important reference for store location selection and other needs.
[0017] According to the above technical solution, in the specific statistical modeling in S2, the store geographical location environment theme score is affected by the number, intensity and degree of clustering of theme points of interest around the store. Taking the POIs around the store as samples, the following model is established for the distribution of theme POIs around the store:
[0018] in:
[0019] s is the store superscript;
[0020] t is the theme subscript, including eight categories: transportation, housing, shopping, food, travel, life, work, and entertainment;
[0021] The distribution of themed POIs around the store;
[0022] It is the theme POI intensity model around the store, which follows the Bernoulli distribution;
[0023] is the probability parameter of the store surrounding theme POI intensity model, which conforms to the Beta distribution;
[0024] α t ,β t is the intensity parameter distribution model variable;
[0025] It is a theme POI aggregation model around the store, which obeys the two-dimensional normal distribution;
[0026] The variance of the theme POI aggregation model around the store conforms to the Gaussian distribution and satisfies is the variance of POI longitude or latitude;
[0027] μ t ,λ t is the aggregation parameter distribution model variable.
[0028] According to the above technical solution, in S2, the model-based calculation process includes input, output and calculation process, as follows:
[0029] Input: t, evaluation topics, including transportation, housing, shopping, food, travel, life, work, and entertainment, a total of eight categories;
[0030] m: total number of evaluated stores;
[0031] The number of POIs found in the t-theme s store;
[0032] tThe number of POIs found in the store of theme s;
[0033] tThe initial probability parameter of the POI intensity model of the theme s store;
[0034] The standard deviation of the distance between all POIs and stores in the theme s store;
[0035] The coordinates of the POI point i near the store s relative to the store;
[0036] α, β: The inter-store strength model follows a Beta distribution a priori, and α, β are model variables;
[0037] μ, λ: The priori model of store aggregation follows a Gaussian distribution, and μ, λ are model variables;
[0038] Output: The scores of s stores under the theme t.
[0039] According to the above technical solution, the calculation process in S2 is as follows:
[0040] 1) Execute in each t;
[0041] 2) The cycle starts;
[0042] 3) In the intensity model, the intensity of thematic POIs around the store follows a Bernoulli distribution, and its probability parameter follows a Beta distribution. The joint probability density function of the intensity model is:
[0043] 4) According to the maximum likelihood estimation, let The partial derivative with respect to p is 0, so find its p. t , where the probability parameter of the intensity model of each store is
[0044] 5) In the aggregation model, the theme POIs around the store obey the two-dimensional normal distribution, the variance of the POI aggregation model conforms to the Gaussian distribution, and the joint probability density function of the aggregation is:
[0045] 6) According to the maximum likelihood estimation, let The partial derivative with respect to σ is 0, so find its σ t , where the standard deviation of the concentration model for each store is
[0046] 7) The probability parameter of the topic intensity model of each store t topic is obtained as and the standard deviation of the concentration model is
[0047] 8) The loop ends;
[0048] 9) Using store information and real-time weather data as independent variables and customer flow data as dependent variables, a linear regression model was constructed;
[0049] 10) R 2 <-model determination coefficient;
[0050] 11), AIC<-Akaike information criterion;
[0051] 12), BIC<-Bayesian Information Criterion;
[0052] 13) Define the topic score calculation method
[0053] 14) Execute in each t;
[0054] 15) The loop starts;
[0055] 16) Theme score of each store's theme
[0056] 17) Add topic score features to the linear regression model
[0057] 18) After getting the score of joining the topic AIC t and BIC t ;
[0058] 19) The loop ends;
[0059] 20) When each t No significant improvement in effect or AIC t Too large or BIC t When it is too big;
[0060] 21) Loop execution;
[0061] 22) Redefine the topic score calculation method and execute steps 14 to 19;
[0062] 23) Observation AIC t 、BIC t ;
[0063] 24) Until each t The effect is significantly improved and AIC t 、BIC t The values are all reasonable;
[0064] 25) The loop ends;
[0065] 26) Keep the corresponding score of t
[0066] 27) It can be applied to real business scenarios such as brand store location selection and grouping, and provide a reference basis for thematic scoring in each dimension for evaluators to select the optimal geographical environment.
[0067] According to the above technical solution, in S3, the reasonable extraction of the theme feature index SSTI mainly refers to using the maximum likelihood estimation to estimate the theme of each store. and
[0068] Specifically, The larger it is, the greater the number of theme POIs around the store. The larger the size, the higher the concentration of theme POIs around the store. Therefore, the following is defined: To describe the POI distribution under the store theme:
[0069] in: is the probability parameter of the POI intensity model of topic t around store s estimated by MLE;
[0070] is the standard deviation of the POI clustering model of topic t around store s estimated by MLE, store The higher the score, the more obvious the theme characteristics are.
[0071] According to the above technical solution, in S4, the linear regression model evaluation system specifically refers to selecting an indicator that can be used to monitor and measure the store's operating conditions or operating goals, such as customer flow or sales volume, using geographic information indicators and other relevant natural features to establish a model, and reflecting the rationality of the geographic information indicators through the model fit gain;
[0072] Among them, the linear regression model has good linear interpretability and feature compatibility, a complete numerical feature evaluation and inference mechanism, and can simply describe the gain of feature superposition. It is naturally adaptable to the evaluation of large temporal and spatial characteristics of offline stores, and it does not require the prerequisites of training and test sets. Therefore, the linear regression model is used to evaluate the size and positive and negative impact of the independent variable characteristics on the dependent variable.
[0073] According to the above technical solution, in S4, the index for measuring the model fit is defined as the coefficient of determination, which represents the percentage of the total variation of the target variable that can be explained by the regression relationship. The closer the value is to 1, the better the fitting effect. The Akaike Information Criterion AIC and the Bayesian Information Criterion BIC can weigh the complexity of the estimated model and the quality of the model fitting the data. The smaller the value, the better the model effect.
[0074] Store information, real-time weather and 8 types of geographical environment indicators are used as independent variables, customer flow data is used as dependent variables, and a linear regression model is constructed. After the 8 types of geographical environment characteristics are added to the evaluation model one by one, R 2 The traffic index has the most obvious effect, followed by the entertainment index and the tourism index. The explainability of the target variable - passenger flow changes is enhanced. The AIC and BIC of each indicator have decreased, indicating that the model fits well and there is no overfitting.
[0075] According to the above technical solution, in S5, the indicator SSTI is applied to real business scenarios, mainly combining the above test results to provide brands with certain business guidance and reference value in store location selection and store grouping;
[0076] Under other similar conditions, the new store address can preferably be based on locations with convenient transportation and certain entertainment facilities, such as commercial complexes near subway entrances. Tourist destinations can also be considered. Based on the needs of actual scenarios, the store's large spatiotemporal indicators provide commercial users with an objective numerical basis for quantifying the surrounding environment, quickly improving the efficiency of geographic assessment.
[0077] Compared with the prior art, the present invention has the following beneficial effects: the present invention has a scientific and reasonable structure and is safe and convenient to use:
[0078] 1. By quantifying the geographic information and natural environment surrounding the target area into eight numerical dimensions, encompassing transportation, residence, shopping, food, travel, living, work, and entertainment, users can quickly and efficiently understand the state of the surrounding environment. Combining geographic information and relevant natural features as indicator scores, this allows users to assess store operating conditions or needs, providing a reference for real-world scenarios such as store site selection, store grouping, and store management. Furthermore, the eight indicators can be used as predictive features for store operations, using regression models to observe whether the surrounding environment has a significant impact on key indicators such as customer flow, and the magnitude of that impact, providing guidance for business operations.
[0079] At the same time, the use of statistical modeling, an efficient and stable method, can greatly reduce personnel costs and eliminate the influence of human subjective factors. Moreover, large spatiotemporal indicators are not in conflict with expert reviews. Large spatiotemporal indicators can be used to assist experts in making more scientific and objective reviews, facilitating better results and achieving the quantification of geographic environment theme scores with business value and strong explanatory power from geographic information.
[0080] 2. The linear regression model has good linear interpretability and feature compatibility, a complete numerical feature evaluation and inference mechanism, and can simply describe the gain of feature superposition. It is naturally adaptable to evaluating the large temporal and spatial characteristics of offline stores. It does not require the prerequisites of training and test sets, which can more quickly improve the efficiency of geographic assessment. BRIEF DESCRIPTION OF THE DRAWINGS
[0081] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.
[0082] In the attached figure:
[0083] FIG1 is a flow chart of the steps of the evaluation method of the present invention. DETAILED DESCRIPTION
[0084] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.
[0085] Example: As shown in Figure 1, the present invention provides a technical solution, a method for evaluating stores by combining relevant indices of the store's surrounding environment. This method mainly quantifies a geographical environment theme score with business value and strong explanatory power from geographic information. The geographical information and relevant natural features are combined as indicator scores to evaluate the store's operating conditions or ability to meet operating needs, providing a reference basis for store business scenarios.
[0086] The assessment steps include:
[0087] S1. Collect geographic information data around offline stores;
[0088] S2. Conduct statistical modeling of entities surrounding the store;
[0089] S3. Reasonably extract the subject feature index SSTI;
[0090] S4, linear regression model evaluation system;
[0091] S5. Indicator SSTI is applied to real business scenarios.
[0092] Based on the above technical solution, in S1, offline store surrounding geographic information data is collected mainly by using different data sources to obtain geographic information around the target store. This geographic information mainly includes the store's surrounding environment information and related natural features;
[0093] After obtaining geographic information, the facilities near the target store are divided into eight categories: transportation, residence, shopping, food, travel, life, work and entertainment. At the same time, for each theme, the number of theme facilities near the target store and the distance information to the target store are obtained.
[0094] Based on the above technical solution, in S2, statistical modeling of the store surrounding entities is mainly carried out by establishing a topic intensity model and a topic aggregation model, and combining the results of the two models to generate a large spatiotemporal index under the topic;
[0095] This paper mainly uses the data of a certain brand as an example and combines it with a linear regression model to verify the impact of large-scale spatiotemporal indicators on customer flow. Once the verification is passed, the large-scale spatiotemporal indicators can be applied to the brand's real business scenarios to help improve the brand's evaluation efficiency of the store's surrounding environment, and ultimately serve as an important reference for store location selection and other needs.
[0096] Based on the above technical solution, in the specific statistical modeling of S2, the store location environment theme score is affected by the number, intensity, and degree of clustering of theme points of interest around the store. Taking the POIs around the store as a sample, the following model is established for the distribution of theme POIs around the store:
[0097] in:
[0098] s is the store superscript;
[0099] t is the theme subscript, including eight categories: transportation, housing, shopping, food, travel, life, work, and entertainment;
[0100] The distribution of themed POIs around the store;
[0101] It is the theme POI intensity model around the store, which follows the Bernoulli distribution;
[0102] is the probability parameter of the store surrounding theme POI intensity model, which conforms to the Beta distribution;
[0103] α t ,β t is the intensity parameter distribution model variable;
[0104] It is a theme POI aggregation model around the store, which obeys the two-dimensional normal distribution;
[0105] The variance of the theme POI aggregation model around the store conforms to the Gaussian distribution and satisfies is the variance of POI longitude or latitude;
[0106] μ t ,λ t is the aggregation parameter distribution model variable.
[0107] Based on the above technical solution, in S2, the model-based calculation process includes input, output, and calculation process, as follows:
[0108] Input: t, evaluation topics, including transportation, housing, shopping, food, travel, life, work, and entertainment, a total of eight categories;
[0109] m: total number of evaluated stores;
[0110] The number of POIs found in the t-theme s store;
[0111] tThe number of POIs found in the store of theme s;
[0112] tThe initial probability parameter of the POI intensity model of the theme s store;
[0113] The standard deviation of the distance between all POIs and stores in the theme s store;
[0114] The coordinates of the POI point i near the store s relative to the store;
[0115] α, β: The inter-store strength model follows a Beta distribution a priori, and α, β are model variables;
[0116] μ, λ: The priori model of store aggregation follows a Gaussian distribution, and μ, λ are model variables;
[0117] Output: The scores of s stores under the theme t.
[0118] Based on the above technical solution, the calculation process in S2 is as follows:
[0119] 1) Execute in each t;
[0120] 2) The cycle starts;
[0121] 3) In the intensity model, the intensity of thematic POIs around the store follows a Bernoulli distribution, and its probability parameter follows a Beta distribution. The joint probability density function of the intensity model is:
[0122] 4) According to the maximum likelihood estimation, let The partial derivative with respect to p is 0, so find its p. t , where the probability parameter of the intensity model of each store is
[0123] 5) In the aggregation model, the theme POIs around the store obey the two-dimensional normal distribution, the variance of the POI aggregation model conforms to the Gaussian distribution, and the joint probability density function of the aggregation is:
[0124] 6) According to the maximum likelihood estimation, let The partial derivative with respect to σ is 0, so find its σ t , where the standard deviation of the concentration model for each store is
[0125] 7) The probability parameter of the topic intensity model of each store t topic is obtained as and the standard deviation of the concentration model is
[0126] 8) The loop ends;
[0127] 9) Using store information and real-time weather data as independent variables and customer flow data as dependent variables, a linear regression model was constructed;
[0128] 10) R 2 <-model determination coefficient;
[0129] 11), AIC<-Akaike information criterion;
[0130] 12), BIC<-Bayesian Information Criterion;
[0131] 13) Define the topic score calculation method
[0132] 14) Execute in each t;
[0133] 15) The loop starts;
[0134] 16) Theme score of each store's theme
[0135] 17) Add topic score features to the linear regression model
[0136] 18) After getting the score of joining the topic AIC t and BIC t ;
[0137] 19) The loop ends;
[0138] 20) When each t No significant improvement in effect or AIC t Too large or BIC t When it is too big;
[0139] 21) Loop execution;
[0140] 22) Redefine the topic score calculation method and execute steps 14 to 19;
[0141] 23) Observation AIC t 、BIC t ;
[0142] 24) Until each t The effect is significantly improved and AIC t 、BIC t The values are all reasonable;
[0143] 25) The loop ends;
[0144] 26) Keep the corresponding score of t
[0145] 27) It can be applied to real business scenarios such as brand store location selection and grouping, and provide a reference basis for thematic scoring in each dimension for evaluators to select the optimal geographical environment.
[0146] Based on the above technical solution, in S3, the reasonable extraction of topic feature index SSTI mainly refers to the use of maximum likelihood estimation to estimate the and
[0147] Specifically, The larger it is, the greater the number of theme POIs around the store. The larger the size, the higher the concentration of theme POIs around the store. Therefore, the following is defined: To describe the POI distribution under the store theme:
[0148] in: is the probability parameter of the POI intensity model of topic t around store s estimated by MLE;
[0149] is the standard deviation of the POI clustering model of topic t around store s estimated by MLE, store The higher the score, the more obvious the theme characteristics are.
[0150] Based on the above technical solution, in S4, the linear regression model evaluation system specifically refers to selecting an indicator that can be used to monitor and measure store operating conditions or operating goals, such as customer flow or sales volume, using geographic information indicators and other relevant natural features to establish a model, and reflecting the rationality of the geographic information indicators through the model fit gain;
[0151] Among them, the linear regression model has good linear interpretability and feature compatibility, a complete numerical feature evaluation and inference mechanism, and can simply describe the gain of feature superposition. It is naturally adaptable to the evaluation of large temporal and spatial characteristics of offline stores, and it does not require the prerequisites of training and test sets. Therefore, the linear regression model is used to evaluate the size and positive and negative impact of the independent variable characteristics on the dependent variable.
[0152] Based on the above technical solution, in S4, the index for measuring the model fit is defined as the coefficient of determination, which represents the percentage of the total variation of the target variable that can be explained by the regression relationship. The closer the value is to 1, the better the fitting effect. The Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC) can weigh the complexity of the estimated model and the quality of the model fitting the data. The smaller the value, the better the model effect.
[0153] Store information, real-time weather and 8 types of geographical environment indicators are used as independent variables, customer flow data is used as dependent variables, and a linear regression model is constructed. After the 8 types of geographical environment characteristics are added to the evaluation model one by one, R 2The traffic index has the most obvious effect, followed by the entertainment index and the tourism index. The explainability of the target variable - passenger flow changes has been enhanced. The AIC and BIC of each indicator have decreased, indicating that the model has a good fit and no overfitting has occurred.
[0154] Based on the above technical solution, in S5, the SSTI indicator is applied to real-world business scenarios. Combining the above test results, it provides brands with certain business guidance and reference value in store location selection and store grouping.
[0155] Under other similar conditions, the new store address can preferably be based on locations with convenient transportation and certain entertainment facilities, such as commercial complexes near subway entrances. Tourist destinations and other places can also be considered. Based on the needs of actual scenarios, the store's large spatiotemporal indicators provide commercial users with an objective numerical basis for quantifying the surrounding environment, quickly improving the efficiency of geographic assessment.
[0156] Finally, it should be noted that the above description is merely 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, those skilled in the art will be able to modify the technical solutions described in the aforementioned embodiments or substitute equivalents for some of the technical features. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.
Claims
1. A method for evaluating a store in combination with relevant indexes of the store's surrounding environment, characterized in that: It mainly quantifies geographic environment theme scores with business value and strong explanatory power from geographic information, combines geographic information and related natural features as indicator scores, evaluates the store's operating conditions or operating needs, and provides a reference for store business scenarios; The evaluation steps include: S1. Collect geographic information data around offline stores; S2. Conduct statistical modeling of entities around the store; S3. Reasonably extract the subject feature index SSTI; S4, linear regression model evaluation system; S5. Indicator SSTI is applied in real business scenarios.
2. The method for evaluating a store in combination with relevant indexes of the store's surrounding environment according to claim 1, characterized in that: In S1, offline store periphery geographic information data is collected mainly by using different data sources to obtain geographic information around the target store, and the geographic information mainly includes store periphery environment information and relevant natural features; After obtaining geographic information, the facilities near the target store are divided into eight categories: transportation, residence, shopping, food, travel, life, work and entertainment. At the same time, for each theme, the number of facilities of that theme near the target store and the distance information to the target store are obtained.
3. The method for evaluating a store in combination with the relevant index of the store's surrounding environment according to claim 1, characterized in that: In S2, statistical modeling of the subject around the store is mainly carried out by establishing a topic intensity model and a topic aggregation model, and combining the results of the two models to generate a large spatiotemporal index under the topic; The linear regression model is mainly used to verify the impact of large-scale spatiotemporal indicators on customer flow. Once verified, the large-scale spatiotemporal indicators can be applied to the brand's real business scenarios to help improve the brand's evaluation efficiency of the store's surrounding environment, and ultimately serve as an important reference for store site selection and other needs.
4. The method for evaluating a store in combination with the relevant index of the store's surrounding environment according to claim 3, characterized in that: In the specific statistical modeling in S2, the store geographical location environment theme score is affected by the number, intensity and degree of aggregation of thematic points of interest around the store. Taking the POI around the store as a sample, the following model is established for the distribution of thematic POI around the store: in: s is the store superscript; t is the theme subscript, including eight categories: transportation, residence, shopping, food, travel, life, work, and entertainment; The distribution of themed POIs around the store; It is the store surrounding theme POI intensity model, which follows Bernoulli distribution; is the probability parameter of the store surrounding theme POI intensity model, which conforms to the Beta distribution; α t ,β t is the strength parameter distribution model variable; It is a theme POI aggregation model around the store, which obeys the two-dimensional normal distribution; The variance of the clustering model of thematic POIs around the store conforms to the Gaussian distribution and satisfies is the variance of POI longitude or latitude; μ t ,λ t is the aggregation parameter distribution model variable.
5. The method for evaluating a store in combination with the relevant index of the store's surrounding environment according to claim 3, characterized in that: In S2, the model-based calculation process includes input, output, and calculation process, as follows: Input: t, evaluation topics, including transportation, residence, shopping, food, travel, life, work, and entertainment; m: total number of evaluated stores; tThe number of POIs found in the store of theme s; tThe number of POIs found by the theme s store; t Initial probability parameters of the POI intensity model of the theme s store; The standard deviation of the distance between all POIs and stores in the theme s store; s The coordinates of the POI point i near the store relative to the store; α, β: The inter-store strength model a priori follows the Beta distribution, α, β are model variables; μ, λ: The a priori Gaussian distribution of the inter-store aggregation model, μ, λ are model variables; Output: The scores of s stores under t theme.
6. The method for evaluating a store in combination with the relevant index of the store's surrounding environment according to claim 5, characterized in that: The calculation process in S2 is as follows: 1) Execute in each t; 2) The cycle starts; 3) In the intensity model, the intensity of the theme POI around the store follows the Bernoulli distribution, and its probability parameter follows the Beta distribution. The joint probability density function of the intensity model is: 4) According to the maximum likelihood estimation, let The partial derivative with respect to p is 0, so p is p t , where the probability parameter of the intensity model for each store is 5) In the aggregation model, the theme POIs around the store obey the two-dimensional normal distribution, the variance of the POI aggregation model conforms to the Gaussian distribution, and the joint probability density function of the aggregation is: 6) According to the maximum likelihood estimation, let The partial derivative with respect to σ is 0, so σ is σ t , where the standard deviation of the concentration model for each store is 7) The probability parameter of the topic intensity model of each store t topic is obtained as and the concentration model standard deviation is 8) The loop ends; 9) Store information and real-time weather data are used as independent variables, and customer flow data is used as the dependent variable to build a linear regression model; 10) R 2 <-model determination coefficient; 11), AIC<-Akaike information criterion; 12), BIC<-Bayesian Information Criterion; 13) Define the topic score calculation method 14) Execute in each t; 15) The cycle starts; 16) Theme score of each store's theme 17) Add topic score features to the linear regression model 18) After getting the score of joining the topic AIC t and BIC t ; 19) The loop ends; 20) When each t No significant improvement in effect or AIC t Too large or BIC t When it is too big; 21) Loop execution; 22) Redefine the topic score calculation method and execute steps 14 to 19; 23) Observation AIC t 、BIC t ; 24) Until each t The effect is significantly improved and AIC t 、BIC t The values are all reasonable; 25) The loop ends; 26) Keep the corresponding score of t 27)、 It can be applied to real business scenarios such as brand store site selection and grouping, and provide a reference basis for thematic scoring in various dimensions for evaluators to choose the optimal geographical environment.
7. The method for evaluating a store in combination with the relevant index of the store's surrounding environment according to claim 1, characterized in that: In S3, the reasonable extraction of the topic feature index SSTI mainly refers to using the maximum likelihood estimation to estimate the and Specifically, The larger it is, the greater the number of themed POIs around the store. The larger the size, the higher the concentration of theme POIs around the store. Therefore, the following To describe the POI distribution under the store theme: in: is the probability parameter of the POI intensity model of topic t around store s estimated by MLE; is the standard deviation of the POI clustering model of topic t around store s estimated by MLE, store The higher the score, the more obvious the theme characteristics are.
8. The method for evaluating a store in combination with relevant indexes of the store's surrounding environment according to claim 1, characterized in that: In S4, the linear regression model evaluation system specifically refers to selecting an indicator that can be used for monitoring and can measure the store's operating conditions or operating goals, and then using geographic information indicators and other relevant natural features to establish a model, and reflecting the rationality of the geographic information indicators through the model fit gain; Among them, the linear regression model has good linear interpretability and feature compatibility, has a complete numerical feature evaluation and inference mechanism, can simply describe the gain of feature superposition, is naturally adaptable to the evaluation of large spatiotemporal characteristics of offline stores, and does not require the prerequisite of training and test sets. Therefore, the linear regression model is used to evaluate the size and positive and negative impact of the independent variable characteristics on the dependent variable.
9. The method for evaluating a store in combination with the relevant index of the store's surrounding environment according to claim 8, characterized in that: In S4, the index for measuring the model fit is defined as the coefficient of determination, which indicates the percentage of the total variation of the target variable that can be explained by the regression relationship. The closer the value is to 1, the better the fitting effect. The Akaike information criterion AIC and the Bayesian information criterion BIC can weigh the complexity of the estimated model and the goodness of the model fitting the data. The smaller the value, the better the model effect. Store information, real-time weather and 8 types of geographical environment indicators were used as independent variables, and customer flow data was used as the dependent variable to build a linear regression model. After adding the 8 types of geographical environment characteristics into the evaluation model one by one, R 2 The traffic index has been improved, among which the effect is the most obvious, followed by the entertainment index and the tourism index. The explainability of the target variable-passenger flow changes is enhanced, and the AIC and BIC of each indicator have decreased, indicating that the model fits well without overfitting.
10. The method for evaluating a store in combination with relevant indexes of the store's surrounding environment according to claim 1, characterized in that: In S5, the indicator SSTI is applied to real business scenarios, mainly combining the above test results to provide brands with certain business guidance and reference value in store location selection and store grouping; Under other similar conditions, the new store address can preferably be located in a place with convenient transportation and certain entertainment facilities. According to the needs of actual scenarios, the store's large spatiotemporal indicators can provide commercial users with an objective numerical basis for quantifying the surrounding environment, thereby quickly improving the efficiency of geographic assessment.
