Shopping mall booth rent evaluation method

Abstract

The invention provides a shopping mall booth rent evaluation method. The method comprises the following steps: step 1, primarily selecting a first-level index and a second-level index; step 2, obtaining a pre-processed secondary index; 3, selecting a supervised learning algorithm suitable for each project stage; step 4, obtaining a data set # imgabs0 # suitable for each project stage; step 5, obtaining a final secondary index; step 6, screening out a data set # imgabs2 # suitable for each project stage from the data sets # imgabs1 #; 7, training the model; 8, embedding the polynomial regression prediction model, the random forest prediction model SK and the random forest prediction model SC into the system; step 9, inputting secondary indexes in the data set # imgabs3 # into a control module of the system; and step 10, obtaining rent evaluation of the booth. According to the stages in the project cycle, the learning supervision model matched with each stage is involved, the factors influencing the rent in each stage are fully mined, and finally the rent of the shopping mall or the booth is accurately evaluated.

Description

Technical Field

[0001] The present invention belongs to the technical field of rent valuation, and particularly relates to a method for valuing the rent of a mall booth. Background Art

[0002] The establishment of a mall rent pricing prediction model aims to provide scientific, accurate, and forward-looking rent pricing strategy support for mall operators. Through in-depth analysis and mining of multi-dimensional information such as a large amount of historical data, market dynamics, tenant characteristics, and the surrounding business environment, it is possible to more accurately predict the reasonable rent levels of various areas and different business types of shops in the mall at different future time periods.

[0003] The core purpose of this model is to help the mall achieve a balance between maximizing revenue and tenant stability. On the one hand, accurate rent pricing can ensure that the mall fully realizes its commercial value, obtains the optimal economic return, and guarantees the continuous development and profit growth of the mall. On the other hand, a reasonable rent level helps to attract high-quality tenants to settle in, maintain the long-term business willingness of tenants, promote the enrichment and optimization of business types in the mall, and enhance the overall business atmosphere and consumer experience.

[0004] In the existing technology, the mall rent pricing methods mainly rely on subjective pricing methods of personnel such as expert scoring, self-cost pricing method, investment income analysis method, and market comparison method, etc., which are mainly financial calculations. There is no relevant mall rent pricing system for reference. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for valuing the rent of a mall booth, which involves a learning and supervision model adapted to each stage according to the stages in the project cycle, fully excavates the factors affecting rent in each stage, and finally accurately evaluates the rent of the mall land plot or booth through the learning and supervision model. The technical solutions adopted are as follows: A method for valuing the rent of a mall booth, comprising the following steps: Step 1: According to the project stages in the project cycle, initially select the first-level indicators and their subordinate second-level indicators involved in each stage of the project cycle; Among them, the project stages include the site selection stage, the opening stage, and the mature mall stage; The first-level indicators in the opening stage include the first-level indicators in the site selection stage, and the first-level indicators in the mature mall stage include the first-level indicators in the opening stage; Step 2: Perform derivative preprocessing on the selected second-level indicators to obtain the preprocessed second-level indicators; Step 3: Select the supervised learning algorithm applicable to each project stage; Among them, the polynomial regression algorithm is selected for the site selection stage; the random forest algorithm is selected for both the opening stage and the mature mall stage; Step 4: Obtain the data sets applicable to each project phase , including the data of secondary indicators and rent information; i = 1~3, corresponding to the site selection phase, the opening phase, and the mature shopping mall phase in sequence; Step 5: Screen the preprocessed secondary indicators in Step 2 to obtain the final secondary indicators; Step 5A: Use the data set corresponding to the mature shopping mall project as the input to train the random forest algorithm in Step 2 to obtain the importance of each secondary indicator; At this time, a random forest prediction model S0 is formed; Step 5B: For the same primary indicator, use the Pearson correlation coefficient to calculate the importance correlation coefficient between its subordinate pairwise secondary indicators. If the importance correlation coefficient ≥ 0.6, it is defined as strongly correlated, and then discard one of the secondary indicators involved in the strong correlation; Step 5C: Aggregate the remaining secondary indicators in the data set , which are the final secondary indicators; Step 6: Compare the final secondary indicators with the characteristics of each project phase, and screen out the data sets applicable to each project phase in the data set , including the data of the final secondary indicators and rent information; Step 7: Use the data set as the input and input it into the polynomial regression algorithm for training to form a polynomial regression prediction model; Use the data set as the input and input it into the random forest prediction model S0 or the random forest algorithm for training to form a random forest prediction model SK; Use the data set as the input and input it into the random forest algorithm or the random forest prediction model S0 for training to form a random forest prediction model SC; In this step, the random forest algorithm has the same structure as the random forest algorithm before the input in Step 5A. Step 8: Embed the polynomial regression prediction model, the random forest prediction model SK, and the random forest prediction model SC into the system;

[0006] Step 9: Input the secondary indicators in the data set into the control module of the system and update the secondary indicator data in real time; Step 10: For the shopping mall land plot to be evaluated, start the polynomial regression prediction model to obtain the rent estimate of the land plot; Step 10: For the shopping mall land plot to be evaluated, start the polynomial regression prediction model to obtain the rent estimate of the land plot; For the booth to be evaluated, starting the random forest prediction model SK or the random forest prediction model SC can obtain the rent estimate of the booth.

[0007] Preferably, the primary indicators initially screened in step 1 specifically include: The primary indicators in the site selection stage include 3 categories of primary indicators: macroeconomic indicators, location indicators, and competitor indicators; The primary indicators in the opening stage include 4 categories of primary indicators: macroeconomic indicators, location indicators, competitor indicators, and mall indicators; The primary indicators in the mature mall stage include 5 categories of primary indicators: macroeconomic indicators, location indicators, competitor indicators, mall indicators, and operation indicators; Among them, the macroeconomic indicators and location indicators are obtained from the network; the competitor indicators are entered by the regional person in charge; The mall indicators and operation indicators are obtained through the mall's internal system.

[0008] Preferably, the macroeconomic indicators are obtained from the official website of the National Bureau of Statistics, and the location indicators are obtained from Anjuke and Baidu Maps.

[0009] Preferably, in step 2, the number of preprocessed secondary indicators corresponding to the site selection stage, opening stage, and mature mall stage are 37, 74, and 137 respectively; Among them, in the site selection stage, the number of secondary indicators under the macroeconomic indicators, location indicators, and competitor indicators are 19, 11, and 7 respectively; In the opening stage, the number of secondary indicators under the macroeconomic indicators, location indicators, competitor indicators, and mall indicators are 19, 11, 7, and 37 respectively; For the primary indicators in the mature mall stage, the number of secondary indicators under the macroeconomic indicators, location indicators, competitor indicators, mall indicators, and operation indicators are 19, 11, 7, 37, and 63 respectively.

[0010] Preferably, in step 5, the number of preprocessed secondary indicators corresponding to the site selection stage, opening stage, and mature mall stage are 13, 25, and 33 respectively; Among them, in the site selection stage, the number of secondary indicators under the macroeconomic indicators, location indicators, and competitor indicators are 5, 3, and 5 respectively; In the opening stage, the number of secondary indicators under the macroeconomic indicators, location indicators, competitor indicators, and mall indicators are 5, 3, 5, and 12 respectively; For the primary indicators in the mature mall stage, the number of secondary indicators under the macroeconomic indicators, location indicators, competitor indicators, mall indicators, and operation indicators are 5, 3, 5, 14, and 6 respectively. Among them, compared with the mall indicators in the opening stage and those in the mature mall stage, 2 secondary indicators are excluded.

[0011] Preferably, the training process of the polynomial regression algorithm in step 7 specifically includes: Step 71: Use the cross-validation method to determine the order of the polynomial regression; Step 72: Use the order determined in step 1 to construct a multi-order polynomial regression model; Step 73: Calculate the coefficient of determination R-squared and the mean squared error regression loss; Step 74: Determine the optimal model and save it.

[0012] Preferably, the training process of the random forest prediction model in step 7 specifically includes: Use MAPE as the evaluation index to evaluate the prediction results of the model; Use the automatic random parameter tuning method to construct a random parameter space, then randomly combine the parameters, and finally find the best combination of random parameters through grid parameter search.

[0013] Compared with the prior art, the advantages of the present invention are: according to the stages in the project cycle, a learning supervision model adapted to each stage is involved, and the factors affecting rent in each stage are fully explored, and finally the rent of the mall or booth is accurately evaluated through the learning supervision model. Specifically: 1. In the new store location selection stage, the operator inputs the target address and surrounding factor indicators into the control module of the system. The display screen of the system (AI rent pricing assistant) embedded with the polynomial regression algorithm model will display the basic data situation and the scoring situation of each dimension of the target plot.

[0014] Normally, the operator only needs to click the start button to activate the polynomial regression prediction model to obtain the estimated average rent of the mall after the project is implemented.

[0015] However, in actual implementation, considering cost and data accuracy issues, in the location selection scenario, in cities where there have been no previous projects, some indicators may be missing or the data may be inaccurate. In this case, new operators need to input them themselves. The 4 indicators to be modified are the regional second-hand housing price, the average second-hand housing price within 5 kilometers around, the number of communities within 5 kilometers around, and the number of newly planned completed housing projects within the next 2 years within 5 kilometers around.

[0016] Therefore, at this time, the operator needs to input the values of the above 4 secondary indicators into the polynomial regression prediction model, and the polynomial regression prediction model calculates the estimated average rent of the mall after the project is implemented.

[0017] Finally, the control module of the system comprehensively analyzes based on the selected multiple target addresses, and the control module of the system gives a site selection recommendation.

[0018] 2. During the opening stage after the mall is built, the control module of the system calculates the predicted layout of the mall's product categories and brands based on the determined mall address, mall basic information, mall map, and other information.

[0019] The operator only needs to click the start button to activate the random forest prediction model SK, and the random forest prediction model SK can calculate the predicted rent situation of each booth based on this layout.

[0020] During the mature mall stage after the mall is built, the control module of the system calculates the actual layout of the mall's product categories and brands based on the determined mall address, mall basic information, mall map, and other information.

[0021] The operator only needs to click the start button to activate the random forest prediction model SC, and the random forest prediction model SC can calculate the predicted rent situation of each booth based on this layout.

[0022] 3. Booth recommendation: The operator inputs data such as target category, brand attributes, expected area, expected rent, and dealer information into the control module of the system through the display screen. The control module of the system will recommend a series of booths and display them on the display screen, and the control module of the system scores the booths. Description of the Drawings

[0023] Figure 1 It is a flowchart of the mall booth rent valuation method; Figure 2 It is a functional display diagram of the AI rent pricing assistant in the new store site selection stage; Figure 3 It is a functional schematic diagram of the AI rent pricing assistant predicting the booth rent in the opening stage or mature mall stage; Figure 4 It is a functional schematic diagram of the AI rent pricing assistant recommending booths and predicting the booth rent in the opening stage or mature mall stage. Detailed Embodiment

[0024] The mall booth rent valuation method of the present invention will be described in more detail below with reference to the schematic diagrams, which show the preferred embodiments of the present invention. It should be understood that those skilled in the art can modify the present invention described herein while still achieving the beneficial effects of the present invention. Therefore, the following description should be understood as a broad guidance for those skilled in the art and not as a limitation to the present invention.

[0025] As Figures 1 - 4 , a mall booth rent valuation method includes the following steps: Step 1: According to the project phases in the project cycle, preliminarily select the first-level indicators and their subordinate second-level indicators involved in each phase of the project cycle. Among them, the project phases include the site selection phase, the opening phase, and the mature mall phase.

[0026] The "mature mall phase" refers to the situation where the mall has been operating normally for more than 1 year and the occupancy rate is higher than 80%.

[0027] In the site selection phase, there is no actual mall.

[0028] As shown in Table 1, the first-level indicators in the opening phase include the first-level indicators in the site selection phase, and the first-level indicators in the mature mall phase include the first-level indicators in the opening phase.

[0029] The first-level indicators preliminarily selected in Step 1 specifically include: The first-level indicators in the site selection phase include 3 categories of first-level indicators: macroeconomic indicators, location indicators, and competitor indicators; The first-level indicators in the opening phase include 4 categories of first-level indicators: macroeconomic indicators, location indicators, competitor indicators, and mall indicators; The first-level indicators in the mature mall phase include 5 categories of first-level indicators: macroeconomic indicators, location indicators, competitor indicators, mall indicators, and operation indicators.

[0030] Among them, the macroeconomic indicators and location indicators are both obtained through the network; the competitor indicators are entered by the regional responsible person; the mall indicators and operation indicators are both obtained through the mall internal system.

[0031] Specifically, the macroeconomic indicators are obtained from the official website of the National Bureau of Statistics, and the location indicators are obtained from Anjuke and Baidu Maps.

[0032] Table 1 The framework diagram of the preliminarily selected indicators Indicator Classification Primarily screened original indicators (selected according to engineers' experience) Macroeconomic Indicators City GDP, City Resident Population, Total Retail Sales of Consumer Goods in the City, Per Capita Disposable Income in the City, Second - hand Housing Prices in the City, Construction Area of Commercial Housing, Sales Area of Commercial Housing, City Consumer Confidence Index, Regional GDP, Regional Resident Population, Total Retail Sales of Consumer Goods in the Region, Per Capita Disposable Income in the Region, Second - hand Housing Prices in the Region (13 indicators) Location Indicators Average Second - hand Housing Price within 5 km, Average Second - hand Housing Price within 3 km, Number of Communities within 5 km, Number of Communities within 3 km, Number of Newly Completed Housing Projects Planned within the Next 2 Years within 5 km, Number of Newly Completed Housing Projects Planned within the Next 2 Years within 3 km, Location Planning (within the development direction of a mature city, no development plan within the next 3 years, key development area planned within the next 3 years), Traffic Conditions (whether there is a main road, whether there is an elevated exit) (8 indicators) Competitor Indicators Number of Competitors, Ranking of This Brand among Competitors, Proportion of Class A Brands of the Competitor with the Highest Ranking, Proportion of Imported Brands of the Competitor with the Highest Ranking, Mall Area of the Competitor with the Highest Ranking, Number of Local Competitors, Number of Chain Competitors (7 indicators) Mall Indicators Mall Area, Opening Time, Mall Type, Mall Positioning, City Level where the Mall is Located, Number of Floors, Number of Parking Spaces, Number of Parking Garage Entrances, Number of Parking Garage Exits, Total Number of Brands in the Mall, Total Number of Series in the Mall, Number of Class A Brands, Occupied Area of Class A Brands, Proportion of the Number of Class A Brands, Proportion of the Occupied Area of Class A Brands, Number of Imported Brands, Occupied Area of Imported Brands, Proportion of the Number of Imported Brands, Proportion of the Occupied Area of Imported Brands, Number of Class B Brands, Occupied Area of Class B Brands, Proportion of the Number of Class B Brands, Proportion of the Occupied Area of Class B Brands, Booth Area, Booth Floor, Booth Grade, Nearest Distance from the Booth to the Elevator, Second Nearest Distance from the Booth to the Elevator, Passenger Flow Grade of the Nearest Elevator, Passenger Flow Grade of the Second Nearest Elevator, Number of Booth Entrances (31 indicators) Business Indicators Number of Purchase Orders, Number of Purchasing Customers, Average Transaction Price per Customer, Total Per Capita Consumption Amount, Number of Purchase Orders for Class A Brands, Number of Purchasing Customers for Class A Brands, Purchase Amount of Class A Brands, Number of Purchase Orders for Imported Brands, Number of Purchasing Customers for Imported Brands, Purchase Amount of Imported Brands, Number of Purchase Orders for Class B Brands, Purchase Amount of Class B Brands (12 indicators) Among them, the competitor refers to other home furnishing store brands that have a competitive relationship in operation; Brand A refers to the average number of shopping malls entered by brands whose number of shopping malls entered ≥ 15% of the number of shopping malls entered by brands in the top category (and the number of shopping malls entered is greater than 15).

[0033] The mall positioning refers to Store No. 1, brand store, and benchmark store.

[0034] The booth level is finally determined by the management personnel considering various indicators.

[0035] The shortest distance from the booth to the elevator is calculated based on the map. The walking distance from the center point of the booth to each elevator on this floor through the passage is calculated, and then the minimum distance is selected.

[0036] The elevator passenger flow level is divided into five levels: 1, 2, 3, 4, and 5. It is sorted according to the actual passenger flow in the mall, and each staircase is assigned independently.

[0037] Step 2: Perform derivative preprocessing on the selected secondary indicators to obtain the preprocessed secondary indicators.

[0038] As shown in Table 2, the number of preprocessed secondary indicators corresponding to the site selection stage, opening stage, and mature mall stage are 37, 74, and 137 respectively. Among them, in the site selection stage, the number of secondary indicators under the macroeconomic indicators, location indicators, and competitor indicators are 19, 11, and 7 respectively. In the opening stage, the number of secondary indicators under the macroeconomic indicators, location indicators, competitor indicators, and mall indicators are 19, 11, 7, and 37 respectively. For the first-level indicators in the mature mall stage, the number of secondary indicators under the macroeconomic indicators, location indicators, competitor indicators, mall indicators, and operation indicators are 19, 11, 7, 37, and 63 respectively.

[0039] Table 2 Index architecture diagram after preprocessing Indicator Classification Pre - processed Secondary Indicators Macroeconomic Indicators City GDP, City Resident Population, Total Retail Sales of Consumer Goods in the City, Per Capita Disposable Income in the City, Second - hand Housing Prices in the City, Construction Area of Commercial Housing, Sales Area of Commercial Housing, City Consumer Confidence Index, Regional GDP, Regional Resident Population, Total Retail Sales of Consumer Goods in the Region, Per Capita Disposable Income in the Region, Second - hand Housing Prices in the Region, Per Capita GDP in the City, Per Capita Total Retail Sales of Consumer Goods in the City, Total Disposable Income in the City, Per Capita GDP in the Region, Per Capita Total Retail Sales of Consumer Goods in the Region, Total Disposable Income in the Region (19 indicators) Location Indicators Average second-hand housing price within 5 km, average second-hand housing price within 3 km, number of residential communities within 5 km, number of residential communities within 3 km, number of new residential projects planned to be delivered within 5 km in the next 2 years, number of new residential projects planned to be delivered within 3 km in the next 2 years, whether it is in the mature urban development direction, whether there is no development plan within the next 3 years, whether it is a key development area planned for the next 3 years, whether there is a main road, whether there is an elevated exit (11 indicators) Competitive product indicators Number of competitors, ranking of the brand among competitors, proportion of Class A brands among the top-ranked competitors, proportion of imported brands among the top-ranked competitors, area of ​​shopping malls of the top-ranked competitors, number of local competitors, number of chain competitors (7 indicators) Market indicators Shopping mall area, opening time, month since opening, whether it is a self-operated shopping mall, whether it is a managed shopping mall, whether it is the No. 1 store, whether it is a benchmark store, whether it is a supreme mall, whether it is a key store, whether it is a general shopping mall, the level of the city, the number of floors, the number of parking spaces, the number of parking garage entrances, the number of parking garage exits, the total number of brands settled in the shopping mall, the total number of series settled in the shopping mall, the number of Class A brands, the settled area of ​​Class A brands, the proportion of the number of Class A brands, the proportion of the settled area of ​​Class A brands, the number of imported brands, the settled area of ​​imported brands, the proportion of the number of imported brands, the proportion of the settled area of ​​imported brands, the number of Class B brands, the settled area of ​​Class B brands, the proportion of the number of Class B brands, the proportion of the settled area of ​​Class B brands, the area of ​​booths, the floor of booths, the level of booths, the closest distance from booths to elevators, the second closest distance from booths to elevators, the passenger flow level of the closest elevator, the passenger flow level of the second closest elevator, the number of booth entrances (37 indicators) Business Indicators Number of purchase orders in the past six months, number of purchase customers in the past six months, average customer price in the past six months, per capita consumption amount in the past six months, number of per capita purchase orders in the past six months, number of customers who purchased Category A brands in the past six months, proportion of customers who purchased Category A brands in the past six months, purchase amount of Category A brands in the past six months, proportion of purchase amount of Category A brands in the past six months, average customer price of Category A brands in the past six months, number of customers who purchased imported brands in the past six months, proportion of customers who purchased imported brands in the past six months, purchase amount of imported brands in the past six months, proportion of purchase amount of imported brands in the past six months, average customer price of imported brands in the past six months, number of customers who purchased Category B brands in the past six months, proportion of customers who purchased Category B brands in the past six months, purchase amount of Category B brands in the past six months, proportion of purchase amount of Category B brands in the past six months, average customer price of Category B brands in the past six months, vehicles entering the market in the past six months, the past year..., the past two years... (63 indicators) Step 3: Select the supervised learning algorithm applicable to each project stage.

[0040] During the initial exploration of the project, a large number of models are experimented and verified. Eventually, the polynomial regression algorithm model is used in the site selection stage; the random forest algorithm model is used in both the opening stage and the mature mall stage.

[0041] The advantages of using the polynomial regression algorithm model in the site selection stage and the random forest algorithm model in both the opening stage and the mature mall stage are as follows: The polynomial regression model is the first method to extend linear regression. It fits non-linear relationships by introducing high-order terms of features, has flexibility and good predictability, has loose requirements for the sample size and the number of feature values, and can have excellent prediction performance with a small number of samples and limited feature values. The random forest can handle high-dimensional data and does not require a large amount of feature engineering. Most importantly, it has a good tolerance for outliers and noise and will not be severely affected by a small amount of abnormal data, resulting in a high prediction performance in the current application scenario.

[0042] Step 4: Obtain the data set applicable to each project stage , including data of secondary indicators and rent information; = 1 to 3, corresponding to the site selection stage, opening stage, and mature mall stage in sequence.

[0043] Among them, the rent information corresponding to the site selection stage is the current rent of a normally operating shopping mall. Among them, the normally operating shopping mall is manually selected by experienced personnel, and the occupancy rate of the selected shopping mall is generally about 95%.

[0044] The rent information corresponding to the opening stage is the current rent of the booths in a normally operating shopping mall.

[0045] The rent information corresponding to the mature shopping mall stage is the current rent of the booths in a normally operating shopping mall.

[0046] During the training process of the prediction models for the above three stages, the same shopping mall samples are used.

[0047] Step 5: Screen the secondary indicators preprocessed in Step 2 to obtain the final secondary indicators; Step 5A: Use the dataset corresponding to the mature shopping mall project as the input to train the random forest algorithm in Step 2 to obtain the importance of each secondary indicator; At this time, a random forest prediction model S0 is formed; Step 5B: For the same primary indicator, use the Pearson correlation coefficient to calculate the importance correlation coefficient between its subordinate pairwise secondary indicators. If the importance correlation coefficient ≥ 0.6, it is defined as strongly correlated, and then discard one of the secondary indicators involved in the strong correlation.

[0048] Table 3 Correlation Coefficient Table 0.8-1.0 Very strong correlation 0.6-0.8 Strong correlation 0.4-0.6 Moderately related 0.2-0.4 Weak correlation 0.0-0.2 Very weak or no correlation The reasons for using the Pearson correlation coefficient are as follows: It helps to screen out the features highly correlated with the target variable and remove the irrelevant or redundant features. This can reduce the complexity of the model, improve the training efficiency and the interpretability of the model; detect whether there is a high degree of linear correlation between features, that is, multicollinearity. Multicollinearity may lead to model instability and inaccurate parameter estimation; reducing irrelevant features can reduce the risk of model overfitting. For example, in this model, through correlation analysis, it is found that the urban GDP and regional GDP data are highly correlated. Combining with the index importance index, finally only the urban GDP is retained as a feature.

[0049] Use the random forest to calculate the importance of each indicator to the result variable, and finally comprehensively screen the indicators based on the Pearson coefficient.

[0050] For example, through importance analysis, the importance of the regional GDP indicator is 0.07, and the importance of the regional resident population quantity indicator is 0.08. Combining with the correlation coefficient of these two indicators being 0.84, finally the regional resident population indicator is selected.

[0051] Calculate the correlation coefficients between various indicators according to the Pearson correlation coefficient. If the correlation coefficient exceeds 0.6, it can be considered that there is a correlation, and then comprehensively consider and screen according to the importance; if the correlation coefficient exceeds 0.8, it can be judged as a strong correlation, and only one of the two indicators can be retained; for example, the correlation coefficient between the regional GDP and the regional resident population is 0.84, and these two indicators can be considered strongly correlated.

[0052] Step 5C: Aggregate the data set The remaining secondary indicators in are the final secondary indicators.

[0053] As shown in Table 4, in Step 5, the number of preprocessed secondary indicators corresponding to the site selection stage, the opening stage, and the mature mall stage are 13, 25, and 33 respectively.

[0054] During the training process of the polynomial regression algorithm in Step 7, a mapping relationship is established between 13 secondary indicators and 1 output quantity.

[0055] During the model training process in the opening stage of Step 7, a mapping relationship is established between 25 secondary indicators and 1 output quantity.

[0056] During the model training process in the mature mall stage of Step 7, a mapping relationship is established between 33 secondary indicators and 1 output quantity.

[0057] Among them, in the site selection stage, the number of secondary indicators under the macroeconomic indicators, location indicators, and competitor indicators are 5, 3, and 5 respectively; In the opening stage, the number of secondary indicators under the macroeconomic indicators, location indicators, competitor indicators, and mall indicators are 5, 3, 5, and 12 respectively; For the first-level indicators in the mature mall stage, the number of secondary indicators under the macroeconomic indicators, location indicators, competitor indicators, mall indicators, and operation indicators are 5, 3, 5, 14, and 6 respectively; Among them, under the mall indicators in the opening stage, compared with the mall indicators in the mature mall stage, 2 secondary indicators are discarded.

[0058] Table 4 Final indicator framework diagram

[0059] Step 6: Compare the final secondary indicators with the characteristics of each project stage, and screen out the data sets applicable to each project stage in the data set to obtain the data sets applicable to each project stage , including the data of the final secondary indicators and rent information. That is, obtain the data set for the training process in Step 7.

[0060] For the mall indicators in the opening stage, compared with those in the mature mall stage, the nearest elevator passenger flow level and the second nearest elevator passenger flow level are excluded.

[0061] The reason is that in the opening stage, it is temporarily impossible to determine the passenger flow situation of each elevator in the mall (the passenger flow situation will become fixed after one year of operation), so these two indicators cannot be added.

[0062] That is, the characteristic of the opening stage is that it is impossible to determine the passenger flow situation of each elevator in the mall. Therefore, these two secondary indicators, namely the nearest elevator passenger flow level and the second nearest elevator passenger flow level, are not required.

[0063] Step 7: Use the data set as the input and input it into the polynomial regression algorithm for training to form a polynomial regression prediction model; Use the data set as the input and input it into the random forest algorithm for training to form a random forest prediction model SK; Use the data set as the input and input it into the random forest algorithm for training to form a random forest prediction model SC.

[0064] The training process of the polynomial regression algorithm specifically includes: Step 71: Use the cross-validation method to determine the order of the polynomial regression; Step 72: Use the order determined in Step 1 to construct a multi-order polynomial regression model; Step 73: Calculate the coefficient of determination R-squared and the mean squared error regression loss; Step 74: Determine the optimal model and save it.

[0065] The training process of the random forest prediction model specifically includes: Use MAPE as the evaluation index to evaluate the prediction results of the model; Use the automatic random parameter tuning method to construct a random parameter space, then randomly combine the parameters, and finally find the best parameters of the random combination through grid parameter search.

[0066] Step 8: Embed the polynomial regression prediction model, the random forest prediction model SK, and the random forest prediction model SC into the system. Among them, the technology involved in the embedding belongs to the existing technology.

[0067] Step 9: Input the secondary indicators in the data set into the control module of the system and update the secondary indicator data in real time; Step 10: For the mall land plot to be evaluated, start the polynomial regression prediction model to obtain the rent estimate of the plot; For the booth to be evaluated, starting the random forest prediction model SK or the random forest prediction model SC can obtain the rent estimate of the booth.

[0068] The training process of the polynomial regression algorithm model specifically includes: Step 71: Use the cross-validation method to determine the order of polynomial regression; Step 72: Use the order determined in Step 1 to construct a multi-order polynomial regression model; Step 73: Calculate the coefficient of determination R-squared and the mean squared error regression loss; Step 74: Determine the optimal model and save it.

[0069] The training process of the random forest prediction model S0 specifically includes: Use MAPE (Mean Absolute Percentage Error) as the evaluation index to evaluate the prediction results of the model.

[0070] ; Among them, represents the true value of the th output value; represents the prediction result of the th output value; is the number of output values.

[0071] Use the automatic random hyperparameter tuning method to construct a random hyperparameter space, including the number of trees to build, the method of selecting the maximum features, the maximum depth of the tree, the minimum number of samples required for node splitting, the minimum number of samples in the leaf node, and the sample sampling method.

[0072] Then randomly combine the hyperparameters, and finally search for the best combination of random hyperparameters through grid search.

[0073] In addition, the specific site selection and booth recommendation functions of the control module involve the following solutions: 1. Site selection: Use min-max scaling to standardize the indicators in the scoring (each indicator is processed into a value between 0 and 100, divided by city. For example, if there are 5 districts in a city, the district with the largest data is 100) Residential score = 0.5 * average second-hand housing price index within 5 kilometers around + 0.2 * number of communities within 5 kilometers around + 0.3 * number of newly planned residential buildings to be completed within the next 2 years within 5 kilometers around Traffic score: main road grade * 0.5 + parking convenience grade * 0.3 + whether there is an elevated exit * 0.2 Consumption Score: Regional Per Capita Disposable Income * 0.4 + Urban Per Capita Disposable Income * 0.25 + Number of Communities within 5 km * 0.25 + Regional Resident Population * 0.1 Business Score: Number of Commercial Districts within 5 km * 0.3 + Number of Commercial Districts within 3 km * 0.6 + Number of Communities within 3 km * 0.1 Population: Urban Resident Population * 0.6 + Regional Resident Population * 0.4 Final Score: Residential Score * 0.35 + Transportation Score * 0.1 * Consumption Score + 0.3 * Business Score + 0.25 * Population Score

[0074] 2. Recommendations (The scopes of the product category area and brand level area have been determined during the planning period. These two parameters need to be exactly the same as the booth planning. The main recommendations are area, rent, and floor).

[0075] Recommended Value: Rent Difference Score * 0.5 + Area Difference Score * 0.35 + Booth Grade Score * 0.15 Rent Difference Score = If rent > budget rent, then it is 100 * (1 - (rent - budget rent) / budget rent); if rent < budget rent, then it is 100 * (1 + 3 * (rent - budget rent) / budget rent).

[0076] Area Difference Score = If area > budget area, then it is 100 * (1 - (area - budget area) / budget area); if area < budget area, then it is 100 * (1 + 3 * (area - budget area) / budget area).

[0077] Booth Grade Score = (6 - booth grade) * 20

[0078] There are 5 levels of booth grades, namely 1, 2, 3, 4, 5 (Level 1 is the highest and Level 5 is the lowest). In summary, compared with the traditional pricing method, using the algorithm modeling method for rent pricing can make full use of a large amount of data resources, including market supply and demand, tenant characteristics, economic environment and other aspects of information, conduct accurate analysis and prediction, and provide a more scientific and objective basis for rent pricing. The specific advantages are mainly reflected in the following aspects: 1. Efficiency and accuracy: Traditional rent pricing methods are often inefficient because they mainly rely on manual collection of limited information and subjective judgment. This process is not only time-consuming and laborious but also vulnerable to human factors. For example, when collecting market data, incomplete data may be obtained due to limited scope or improper methods, affecting the accuracy of pricing. In contrast, AI rent pricing, with its powerful data analysis and fast processing capabilities, can integrate a large amount of multi-dimensional data in a short time and conduct in-depth analysis through complex algorithms to quickly obtain pricing results. In terms of accuracy, relying on comprehensive data and scientific algorithms, AI rent pricing can more accurately grasp key factors such as market supply and demand relationships and tenant characteristics, thus giving rent prices that are more in line with the actual market situation.

[0079] 2. Flexibility and adaptability: Traditional rent pricing shows obvious rigidity and sluggishness in the face of market changes and special situations. For example, when the economic situation suddenly changes, policies are adjusted, or major construction projects occur in the area, traditional pricing methods are difficult to respond quickly. It often requires a period of observation and manual re-evaluation to adjust the rent. In contrast, AI rent pricing can monitor market dynamics and changes in various influencing factors in real time and automatically and quickly adjust rent prices through preset algorithms and models. In special situations such as sudden changes in rental demand caused by emergencies, AI rent pricing can immediately respond and flexibly adjust price strategies to adapt to the new market demands.

[0080] 3. Tenant experience and satisfaction: Since traditional rent pricing is difficult to meet the personalized needs of tenants, tenants may feel that the rent plan is not reasonable enough, resulting in low satisfaction. For example, for tenants with different income levels and various rental needs, traditional pricing cannot provide differentiated rent plans. AI rent pricing can provide personalized rent plans based on tenants' personal circumstances such as income, occupation, and rental history, making tenants feel valued and understood, thereby improving tenant satisfaction and loyalty. In addition, AI rent pricing can more accurately reflect market value, making tenants feel that the rent is more fair and reasonable, further enhancing the tenant experience.

[0081] The above are only the preferred embodiments of the present invention and do not impose any limitation on the present invention. Any person skilled in the art within the technical field of the present invention, without departing from the scope of the technical solution of the present invention, makes any form of equivalent replacement or modification and other changes to the technical solution and technical content disclosed by the present invention, which are all within the content of the technical solution of the present invention and still fall within the protection scope of the present invention.

Claims

1. A method for evaluating the rental of a shopping mall booth, characterized in that: The following steps are involved: Step 1: According to the project stages in the project cycle, preliminarily select the primary indicators and subordinate secondary indicators involved in each stage of the project cycle; Among them, the project stages include site selection stage, opening stage and mature shopping mall stage; The first-level indicators of the opening stage include the first-level indicators of the site selection stage, and the first-level indicators of the mature shopping mall stage include the first-level indicators of the opening stage; Step 2: Perform derivative preprocessing on the selected secondary indicators to obtain the preprocessed secondary indicators; Step 3: Select the supervised learning algorithm applicable to each project stage; Among them, the polynomial regression algorithm is used in the site selection stage; the random forest algorithm is used in the opening stage and the mature shopping mall stage; Step 4: Obtain the appropriate dataset for each project stage , including data on secondary indicators and rental information; =1~3, corresponding to the site selection stage, opening stage and mature shopping mall stage; Step 5: Screen the secondary indicators preprocessed in step 2 to obtain the final secondary indicators; Step 5A: Dataset corresponding to mature shopping mall projects As input, the random forest algorithm in step 2 is trained to obtain the importance of each secondary indicator; At this point, the random forest prediction model S0 is formed; Step 5B: For the same primary indicator, use the Pearson correlation coefficient to calculate the importance correlation coefficient between the two secondary indicators under it. If the importance correlation coefficient is ≥ 0.6, it is defined as a strong correlation, and one of the secondary indicators involved in the strong correlation is discarded; Step 5C: Aggregate the dataset The remaining secondary indicators are the final secondary indicators; Step 6: Compare the final secondary indicators with the characteristics of each project stage in the data set Filter out the datasets that are suitable for each project stage , including the final secondary indicator data and rental information; Step 7: Dataset As input, it is input into the polynomial regression algorithm for training to form a polynomial regression prediction model; The dataset As input, it is input into the random forest algorithm for training to form a random forest prediction model SK; The dataset As input, it is input into the random forest algorithm for training to form a random forest prediction model SC; Step 8: embed the polynomial regression prediction model, the random forest prediction model SK and the random forest prediction model SC into the system; Step 9: Dataset The secondary indicators in the system are input into the control module, and the secondary indicator data are updated in real time; Step 10: For the shopping mall land to be evaluated, start the polynomial regression prediction model to obtain the rental estimate of the land; For the booth to be evaluated, the random forest prediction model SK or the random forest prediction model SC is started to obtain the rental estimate of the booth.

2. The method for evaluating the rental price of a shopping mall booth according to claim 1, characterized in that: The primary indicators initially selected in step 1 include: The first-level indicators in the site selection stage include three types of first-level indicators: macroeconomic indicators, location indicators, and competitive product indicators; The first-level indicators at the opening stage include four categories of first-level indicators: macroeconomic indicators, location indicators, competitive product indicators, and shopping mall indicators; The first-level indicators of the mature market stage include five categories of first-level indicators: macroeconomic indicators, location indicators, competitive product indicators, market indicators and operating indicators; Among them, the macroeconomic indicators and location indicators are obtained from the Internet; the competitive product indicators are entered by the regional person in charge; The mall indicators and operating indicators are all obtained through the mall’s internal system.

3. The method for evaluating the rental price of a shopping mall booth according to claim 2, characterized in that: The macroeconomic indicators are obtained from the official website of the National Bureau of Statistics, and the location indicators are obtained from Anjuke and Baidu Maps.

4. The method for evaluating the rental price of a shopping mall booth according to claim 2, characterized in that: In step 2, the number of pre-processed secondary indicators corresponding to the site selection stage, opening stage, and mature shopping mall stage are 37, 74, and 137, respectively; Among them, in the site selection stage, the number of secondary indicators under macroeconomic indicators, location indicators and competitive product indicators are 19, 11 and 7 respectively; In the opening stage, the number of secondary indicators under macroeconomic indicators, location indicators, competitive product indicators and shopping mall indicators are 19, 11, 7 and 37 respectively; The number of secondary indicators under the primary indicators of the mature market stage, namely macroeconomic indicators, location indicators, competitive product indicators, market indicators and operating indicators, are 19, 11, 7, 37 and 63 respectively.

5. The method for evaluating the rental price of a shopping mall booth according to claim 4, characterized in that: In step 5, the number of pre-processed secondary indicators corresponding to the site selection stage, opening stage, and mature shopping mall stage are 13, 25, and 33, respectively; Among them, in the site selection stage, the number of secondary indicators under the macroeconomic indicators, location indicators and competitive product indicators are 5, 3 and 5 respectively; In the opening stage, the number of secondary indicators under macroeconomic indicators, location indicators, competitive product indicators and shopping mall indicators are 5, 3, 5 and 12 respectively; The number of secondary indicators under the primary indicators of the mature market stage, macroeconomic indicators, location indicators, competitive product indicators, market indicators and operating indicators, are 5, 3, 5, 14 and 6 respectively; Among them, the shopping mall indicators in the opening stage have discarded two secondary indicators compared with the shopping mall indicators in the mature shopping mall stage.

6. The method for evaluating the rental price of a shopping mall booth according to claim 1, characterized in that: The training process of the polynomial regression algorithm in step 7 specifically includes: Step 71, using a cross-validation method to determine the order of polynomial regression; Step 72: construct a multi-order polynomial regression model using the order determined in step 1; Step 73, calculate the determination coefficient R square and the mean square error regression loss; Step 74: Determine the optimal model and save it.

7. The method for evaluating the rental price of a shopping mall booth according to claim 1, characterized in that: The training process of the random forest prediction model in step 7 specifically includes: Use MAPE as an evaluation indicator to evaluate the prediction results of the model; The random parameter space is constructed by automatic random parameter adjustment, and then the parameters are randomly combined. Finally, the best parameters of the random combination are found through grid parameter search.

Images