CN114510988A - Method, device, computer equipment and storage medium for building a site location model - Google Patents

Abstract

The present application relates to a method, device, computer equipment, storage medium and computer program product for constructing a site selection model. The method includes: obtaining original feature data corresponding to site selection from channel multi-source data based on a preset feature index system; obtaining evidence weight and information value corresponding to each feature in the original feature data, based on the evidence weight According to the information value, the smart features corresponding to the site selection model are screened; based on the smart features, the site site selection model is constructed through principal component analysis and entropy value method. The present application conducts feature screening by pre-establishing a feature index system, and then selects smart features corresponding to the site selection model from the screened features, thereby constructing the site site selection model through principal component analysis and entropy method. The solution of the present application analyzes the multi-source data of site location selection through multiple feature screening, thereby constructing a site site selection model, which can effectively realize the comprehensive consideration of the site site location process.

Description

Method, device, computer equipment and storage medium for building a site location model

technical field

The present application relates to the field of computer technology, and in particular, to a method, apparatus, computer equipment, storage medium and computer program product for constructing a site selection model.

Background technique

With the development of network technology, big data technology has emerged. Big data, or huge amount of data, refers to the amount of data involved that is too large to be captured and managed within a reasonable time through mainstream software tools. , process, and organize information into a more positive purpose to help business decision-making.

As the main channel for enterprises to provide offline services to the public, the network of enterprises plays an important role in attracting high-quality customers, improving enterprise efficiency, and realizing enterprise development strategies. At the same time, due to the large investment, fixed location and relevant policy requirements, the network resources are also limited. It has also become a choice to solve the problem of site location through big data technology.

In traditional technology, the calculation of site location is generally performed by quantitative analysis methods, including simulation methods and heuristic methods, such as genetic algorithm, particle swarm algorithm, ant colony algorithm, etc. However, these methods focus on the optimal solution of the algorithm, but The consideration is not comprehensive enough, and there are few influencing factors considered, so it is impossible to conduct in-depth analysis of various types of multi-source data.

SUMMARY OF THE INVENTION

Based on this, it is necessary to provide a method, device, computer equipment, computer-readable storage medium and computer program product for constructing a site location model that can analyze multi-source data to achieve comprehensive consideration for the above technical problems.

In a first aspect, the present application provides a method for constructing a site selection model. The method includes:

Based on the preset feature index system, obtain the original feature data corresponding to the site selection from the multi-source data of the channel;

Obtain the evidence weight and the information value corresponding to each feature in the original feature data, and screen the smart features corresponding to the site selection model based on the evidence weight and the information value;

Based on the smart features, a site selection model is constructed by principal component analysis and entropy method.

In one embodiment, based on the preset feature index system, obtaining the original feature data corresponding to the site selection from the channel multi-source data includes:

Obtain the channel multi-source data corresponding to the site selection;

Based on the preset feature index system, the multi-source data of the channel is classified and filtered, and the original feature data corresponding to the site selection is obtained.

In one embodiment, the obtaining of the evidence weight and information value corresponding to each feature in the original feature data, and based on the evidence weight and the information value, screening the smart features corresponding to the site location model includes:

Obtain the null value ratios corresponding to various types of characteristic data in the original characteristic data;

screening the original feature data based on the null rate;

Obtain the evidence weight and information value corresponding to the original feature data that has been screened;

Eliminate the original feature data whose information value is lower than the preset information value threshold, and obtain the first initial feature;

Identify the nonlinear relationship corresponding to the first initial feature based on the evidence weight, and use the first initial feature of the inverted U-shaped relationship corresponding to the nonlinear relationship corresponding to the first initial feature as the second initial feature;

Based on the second initial feature, a smart feature corresponding to the site location model is acquired.

In one embodiment, before acquiring the smart feature corresponding to the site selection model based on the initial feature, the method further includes:

The initial random forest model is trained based on the global search method to obtain the random forest model;

The smart features corresponding to the acquisition of the site location model based on the initial features include:

The initial features are screened based on a random forest model to obtain smart features.

In one of the embodiments, based on the smart features, constructing a site selection model by principal component analysis and entropy value method includes:

Perform dimensionality reduction processing on the smart features through principal component analysis to obtain dimensionality reduction features;

The dimensionality reduction feature is processed by the entropy value method, and the feature weight corresponding to the dimensionality reduction feature is obtained;

Based on the dimension reduction feature and the feature weight corresponding to the dimension reduction feature, a network site selection model is constructed.

In one of the embodiments, the preset feature index system, the channel service dimension index system and the regional development dimension index system, based on the smart features, after the network site selection model is constructed by principal component analysis and entropy value method, further include:

Obtain the site selection data corresponding to each candidate site in the target site selection area;

Input the site location data into the site location model, and obtain the channel service capability score and the regional development potential score corresponding to each candidate location;

Based on the channel service capability score and the regional development potential score, a site selection location corresponding to the target site selection area is obtained.

In a second aspect, the present application also provides an apparatus for constructing a site selection model. The device includes:

The data screening module is used to obtain the original feature data corresponding to the site selection from the multi-source data of the channel based on the preset feature index system;

A feature screening module, configured to obtain the evidence weight and information value corresponding to each feature in the original feature data, and based on the evidence weight and the information value, screen smart features corresponding to the site selection model;

The model building module is used for building a site selection model by principal component analysis and entropy value method based on the smart features.

In a third aspect, the present application also provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

Based on the preset feature index system, obtain the original feature data corresponding to the site selection from the multi-source data of the channel;

Obtain the evidence weight and the information value corresponding to each feature in the original feature data, and screen the smart features corresponding to the site selection model based on the evidence weight and the information value;

Based on the smart features, a site selection model is constructed by principal component analysis and entropy method.

In a fourth aspect, the present application also provides a computer-readable storage medium. The computer-readable storage medium has a computer program stored thereon, and when the computer program is executed by the processor, the following steps are implemented:

Based on the preset feature index system, obtain the original feature data corresponding to the site selection from the multi-source data of the channel;

Obtain the evidence weight and the information value corresponding to each feature in the original feature data, and screen the smart features corresponding to the site selection model based on the evidence weight and the information value;

Based on the smart features, a site selection model is constructed by principal component analysis and entropy method.

In a fifth aspect, the present application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the following steps:

Based on the preset feature index system, obtain the original feature data corresponding to the site selection from the multi-source data of the channel;

Obtain the evidence weight and the information value corresponding to each feature in the original feature data, and screen the smart features corresponding to the site selection model based on the evidence weight and the information value;

Based on the smart features, a site selection model is constructed by principal component analysis and entropy method.

The above-mentioned method, device, computer equipment, storage medium and computer program product for constructing a site selection model, wherein the method is based on a preset feature index system, and obtains original feature data corresponding to site site selection from channel multi-source data; The evidence weight and information value corresponding to the item features, based on the evidence weight and information value, screen the smart features corresponding to the site selection model; based on the smart features, the site site selection model is constructed by principal component analysis and entropy value method. The method for constructing a site selection model of the present application performs feature screening by pre-establishing a feature index system, and then selects smart features corresponding to the site selection model from the screened features, so as to construct a site selection model through principal component analysis and entropy method. address model. The solution of the present application analyzes the multi-source data of site location selection through multiple feature screening, thereby constructing a site site selection model, which can effectively realize the comprehensive consideration of the site site selection process.

Description of drawings

1 is an application environment diagram of a method for constructing a site selection model in one embodiment;

Fig. 2 is a schematic flow chart of a method for constructing a site selection model in one embodiment;

Fig. 3 is a sub-flow schematic diagram of step 201 in Fig. 2 in one embodiment;

4 is a schematic diagram of a data classification and screening process based on multi-source data aggregation and an evaluation index system in one embodiment;

Fig. 5 is a sub-flow schematic diagram of step 203 in Fig. 2 in one embodiment;

Fig. 6 is a sub-flow schematic diagram of step 205 in Fig. 2 in one embodiment;

7 is a schematic flow chart of the steps of site location selection according to a site location model in one embodiment;

8 is a schematic diagram of the classification quadrant corresponding to the site selection score in one embodiment;

9 is a structural block diagram of an apparatus for constructing a site selection model in one embodiment;

Figure 10 is a diagram of the internal structure of a computer device in one embodiment.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present application more clearly understood, the present application will be described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, but not to limit the present application.

The method for constructing a site selection model provided by the embodiment of the present application can be applied to the application environment shown in FIG. 1 . The terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process in the process of building the site selection model. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network server. The terminal 102 may send a site location model construction request to the server 104, and after receiving the site site site model construction request, the server 104 obtains the original feature data corresponding to the site site location from the multi-channel multi-source data based on the preset feature index system; The evidence weight and information value corresponding to each feature in the original feature data, based on the evidence weight and information value, filter the smart features corresponding to the network location model to obtain the evidence weight and information value corresponding to each feature in the original feature data, based on the evidence weight According to the information value, the smart features corresponding to the site selection model are screened; based on the smart features, the site site selection model is constructed by principal component analysis and entropy method. The terminal 102 can be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, IoT devices and portable wearable devices, and the IoT devices can be smart speakers, smart TVs, smart air conditioners, smart vehicle-mounted devices, etc. . The portable wearable device may be a smart watch, a smart bracelet, a head-mounted device, or the like. The server 104 can be implemented by an independent server or a server cluster composed of multiple servers.

In one embodiment, as shown in FIG. 2, a method for constructing a site selection model is provided, and the method is applied to the server 104 in FIG. 1 as an example for description, including the following steps:

Step 201 , based on a preset feature index system, obtain original feature data corresponding to the site selection from channel multi-source data.

Among them, the preset feature index system refers to determining the features that affect the site selection process through various index data, and each site can be analyzed and evaluated through the preset feature index system to obtain the site selection corresponding to the site location. The preset feature index system specifically includes the business circle ecological evaluation system, the transportation convenience evaluation system, the competition distribution evaluation system, the customer value evaluation system and the business demand evaluation system. The site selection may be affected by factors such as regional economic fundamentals, population distribution, market competition, and road traffic. Therefore, when constructing a network site selection model, an index system that affects the site selection can be pre-established by regional economic basic factors, population distribution factors, market competition factors, and road traffic factors. Then, through the preset feature index system, the original feature data corresponding to the site selection is obtained from the multi-source data of the channel. Among them, the basic economic factors refer to the economic level, surrounding business circles, infrastructure and other factors of the site selection area. The higher the economic level of a region, the higher the consumption level of users, and the greater the demand for corporate network services, the more suitable it is as a choice for network layout. Population distribution factors refer to the number of users, age group distribution, purchasing power index, financial characteristics, customer group distribution, etc. within the coverage of a region. As the online download body of enterprise service users, outlets are mainly to serve more customers with better quality as much as possible. The market competition factor means that as the market matures and the choices of users become more diverse, in the same area, if the distribution density of enterprises with similar service levels is relatively large, it will lead to intensified competition and scattered users. development has adverse effects. Road traffic factors refer to factors such as road conditions, the distribution of public transport facilities, and road connections that should be comprehensively considered in the process of site selection. The road traffic conditions in an area determine whether it can maximize the convenience for customers. The specific package library of road traffic factors: 1. Market competition factors, as the market becomes more and more mature, the choices of users also tend to be diversified. If the banking outlets with similar service levels in the same area are densely distributed, it will lead to intensified competition and scattered users, which is not conducive to the long-term development of the industry; 2. Road traffic factors. In the site selection decision, factors such as road construction quality, public transportation convenience, and road connection smoothness should be comprehensively considered. The road traffic conditions in an area determine whether it can provide maximum convenience to customers. The original feature data refers to the feature data filtered from the basic data collected from various channels based on the preset feature index system.

Specifically, in the process of site location selection, it is necessary to obtain feature data for building a model first. However, the sources of data that affect the site selection are complex and diverse, so a preset feature index system corresponding to the site site selection process can be established. Based on the index system, the original feature data that affects the site selection process is collected. Pre-collected network site location GIS (Geographic Information System) data, mobile signaling data, web crawler data, map software API data and other basic data for site location selection, and then based on the preset feature index system for basic data. Perform evaluation and identification to obtain the original feature data corresponding to the site selection.

Step 203: Obtain evidence weight and information value corresponding to each feature in the original feature data, and screen smart features corresponding to the site selection model based on the evidence weight and information value.

Among them, in the weight of evidence and the value of information, the weight of evidence indicates the predictive ability of the independent variable relative to the dependent variable. Information value is used to rank variables according to their importance. Evidence weight and information value are currently generally used as benchmarks for screening variables such as probability of default in credit risk modeling projects. In the present application, the weight of evidence and the value of information are used as benchmarks for screening features in the site selection model. Smart features refer to features that greatly contribute to the site selection process. Based on these smart features, the site site selection process can be optimized, so that a more suitable site can be selected from multiple candidate sites.

Since the features involved in the original feature data are too complex and diverse, there are still many features that have no or too little influence on the site selection process, so after obtaining the original feature data, it is necessary to perform corresponding feature screening to obtain the original The weight of evidence and the value of information corresponding to each feature in the feature data, based on the weight of evidence and the value of information, the smart features corresponding to the site selection model are screened, so as to ensure the accuracy of the site selection model and reduce the complexity of the model. Improve the construction efficiency of the model construction process and the calculation efficiency of the subsequent network site location process. When screening smart features, the contribution of each original feature to the site selection process can be calculated, and then the smart features can be determined based on the ranking of the contribution. In one embodiment, the original features ranked in the top 50% of the contribution may be regarded as smart features.

Step 205 , based on the smart features, construct a site selection model by principal component analysis and entropy method.

Among them, principal component analysis refers to a statistical method. A set of potentially correlated variables is transformed into a set of linearly uncorrelated variables through orthogonal transformation, and the transformed set of variables is called principal components. In many cases, there is a certain correlation between variables. When there is a certain correlation between two variables, it can be explained that the information of the two variables reflecting the subject has a certain overlap. Principal component analysis is to delete redundant variables (closely related variables) for all the variables originally proposed, and establish as few new variables as possible, so that these new variables are uncorrelated, and these new variables reflect The information on the subject should be kept as original as possible. Therefore, after the smart features are obtained, the smart features can be combined through principal component analysis, so as to realize the dimension reduction of the smart features, further reduce the complexity of the analysis process, and improve the computational efficiency. The entropy method is a mathematical method used to judge the degree of dispersion of an indicator. The greater the degree of dispersion, the greater the impact of the index on the comprehensive evaluation. The entropy value can be used to judge the degree of dispersion of an indicator.

Specifically, after obtaining the smart features, the correlation between the smart features can be determined first, and then the correlated smart features can be converted into dimension-reduced features through orthogonal transformation, so as to delete the possible duplicate variables in the smart features, and realize Dimensionality reduction of smart features, and keep the original information as much as possible. After the principal component analysis is completed, each dimension reduction feature can be evaluated by the entropy method, so as to obtain the corresponding weight of the dimension reduction feature, and realize the construction of the network site selection model. After obtaining the site selection model, the corresponding features of each site can be input into the site site model, and the site site model can calculate the corresponding features of these sites according to the summed up dimensionality reduction features and feature weights. The site selection score can be obtained to obtain a better site selection address and realize the optimization of the site site selection process.

The above-mentioned method for constructing an outlet location model, based on a preset feature index system, obtains the original feature data corresponding to the outlet location from the multi-source data of channels; Information value, screen out the smart features corresponding to the site selection model; based on the smart features, the site site selection model is constructed through principal component analysis and entropy value method. The method for constructing a site selection model of the present application performs feature screening by pre-establishing a feature index system, and then selects smart features corresponding to the site selection model from the screened features, so as to construct a site selection model through principal component analysis and entropy method. address model. The solution of the present application analyzes the multi-source data of site location selection through multiple feature screening, thereby constructing a site site selection model, which can effectively realize the comprehensive consideration of the site site selection process.

In one embodiment, as shown in FIG. 3, step 201 includes:

Step 302: Acquire channel multi-source data corresponding to the site selection.

Step 304 , classify and filter the channel multi-source data based on the preset feature index system, and obtain the original feature data corresponding to the site selection.

Among them, the channel multi-source data includes basic data collected from multiple different channels, including the site location GIS data, mobile signaling data, web crawler data, and map software API data corresponding to the existing sites. basic data. These data come from different information channels. By collecting multi-source data from these channels, we can display information about the location selection of existing outlets from different channels such as internal and external channels.

Specifically, when constructing a site location model, it is first necessary to determine characteristic data affecting site location. Therefore, it is possible to build a network location model, and first establish a characteristic index system. Then, based on the preset feature index system, the channel multi-source data is classified and filtered. Referring to FIG. 4, in one of the embodiments, when selecting a location for a bank outlet, the multi-source data of channels is classified into factors such as economic fundamentals, population distribution factors, market competition factors, and road traffic factors that affect the location selection of the outlets. After the data is collected, the corresponding channel multi-source data can be classified and filtered under each preset feature index system to determine the data corresponding to each preset feature index system, so as to construct and obtain the original feature data corresponding to the site selection. In this embodiment, the channel multi-source data is classified and screened based on the preset feature index system, and the channel multi-source data can be effectively obtained from the channel multi-source data for classification and screening, and the corresponding network site selection can be obtained from the channel multi-source data. raw feature data.

In one embodiment, as shown in FIG. 5 , step 203 includes:

Step 502: Obtain the null ratios corresponding to various types of feature data in the original feature data.

Step 504 , filter the original feature data based on the null rate.

Step 506: Obtain the evidence weight and information value corresponding to the original feature data that has been screened.

Step 508: Eliminate the original feature data whose information value is lower than the preset information value threshold, and obtain the first initial feature.

Step 510: Identify the nonlinear relationship corresponding to the first initial feature based on the weight of evidence, and use the first initial feature where the nonlinear relationship corresponding to the first initial feature is an inverted U-shaped relationship as the second initial feature.

Step 512: Based on the second initial feature, acquire the smart feature corresponding to the site location model.

Among them, the null value rate, also known as the field null value rate, refers to the probability that the field value corresponding to each feature in the original feature data is null. In the weight of evidence and the value of information, the weight of evidence indicates the predictive power of the independent variable relative to the dependent variable. Information value is used to rank variables according to their importance. Evidence weight and information value are currently generally used as benchmarks for screening variables such as probability of default in credit risk modeling projects. In this application, the weight of evidence and the value of information are used as benchmarks for screening initial features in the site selection model.

Specifically, after obtaining the original feature data, because the features involved in the original feature data are too complex and diverse, there are still many features that have no or little influence on the site location selection process, so after obtaining the original feature data, Corresponding feature screening is also required to obtain the evidence weight and information value corresponding to each feature in the original feature data, and based on the evidence weight and information value, the smart features corresponding to the site selection model are screened. In this process, the first screening can be performed based on the null value rates corresponding to various types of feature data in the original feature data, and then the second screening can be performed according to the evidence weights and information values corresponding to various types of feature data. When filtering based on the null value rate, the feature data with the null value rate greater than the preset null value rate threshold may be specifically filtered out. The integrity of the feature data with a high null rate cannot be guaranteed. In order to ensure the availability of the site selection model, the feature data with a null rate greater than the preset null rate threshold can be removed first. When performing feature screening based on evidence weight and information value, it is possible to remove features whose information value is lower than the preset information value threshold, such as removing feature data whose information value is less than 0.5, and select features whose evidence weight is an inverted U shape. , as the initial features in the original feature data. Then, based on the initial features, further screening can be performed to obtain smart features for the construction of the site location model. In this embodiment, various types of feature data in the original feature data are screened based on the null value rate, the weight of evidence and the information value, which can effectively screen out the initial features used for the modeling of site selection from the original feature data, thereby ensuring that The accuracy of smart feature recognition.

In one embodiment, before step 510, the method further includes: training an initial random forest model based on a global search method to obtain a random forest model. Step 510 includes: screening initial features based on a random forest model to obtain smart features.

Among them, the global search method refers to conducting a search on the entire feasible set to find extreme points. These methods only need to calculate the objective value of the function and do not need to differentiate the objective function. Random forest refers to a classifier that uses multiple trees to train and predict samples.

Specifically, after the initial features are obtained, further feature screening can be performed. This process can be performed through the random forest algorithm. The model is trained based on the global search method, and the optimal parameter of the random forest is set to n_estimators\max_features\max_depth, and then Using the cross-validation loop calculation, the importance of each initial feature is averaged and the smart features are screened. In a specific embodiment, the corresponding importance of each initial feature can be determined through a random forest model, and then the top 50% of the features are screened out. features as smart features of the site location model. In another embodiment, the lasso algorithm can also be used to filter smart features. In this case, the penalty term can be set to 0.01, and then features with coefficients other than 0 are obtained, and these features are regarded as smart features. In this embodiment, the random forest model is used to screen the initial features, so that more effective smart features can be obtained, thereby improving the success rate of building the site selection model and the accuracy of site site selection.

In one embodiment, as shown in FIG. 6, step 205 includes:

Step 601: Perform dimension reduction processing on smart features through principal component analysis to obtain dimension reduction features.

In step 603, the dimension reduction feature is processed by the entropy value method, and the feature weight corresponding to the dimension reduction feature is obtained.

Step 605 , based on the dimension reduction feature and the feature weight corresponding to the dimension reduction feature, construct a site selection model.

Among them, principal component analysis refers to a statistical method. A set of potentially correlated variables is transformed into a set of linearly uncorrelated variables through orthogonal transformation, and the transformed set of variables is called principal components. In the present application, after obtaining smart features, different smart features can be combined through principal component analysis, thereby realizing dimensionality reduction of smart features, obtaining dimensionality-reducing features, further reducing the complexity of the analysis process, and improving computational efficiency.

The entropy method is a mathematical method used to judge the degree of dispersion of an indicator. The greater the degree of dispersion, the greater the impact of the index on the comprehensive evaluation. The entropy value can be used to judge the degree of dispersion of an indicator. In this application, the entropy value method is mainly used to evaluate various dimensionality reduction features, so as to obtain the corresponding weights of the dimensionality reduction features, so as to realize the construction of the network site location model. After the dimension reduction feature and the weight of the dimension reduction feature are obtained, the site selection model can be constructed based on these two data. The features corresponding to each site location can be input into the site site selection model, and the site site selection model can be Input, get the corresponding dimensionality reduction feature value, and then based on its corresponding feature weight, the site selection score corresponding to these site selection locations can be calculated separately, so as to obtain a better site selection address, and realize the process of site site selection. optimization. In this embodiment, the optimization of the model building process is carried out through PCA combined with the entropy method, which can effectively improve the operation efficiency of the model, thereby improving the processing efficiency of the site location selection process.

In one embodiment, the preset feature index system, channel service dimension index system and regional development dimension index system, as shown in FIG. 7 , after step 205, further includes:

Step 702: Obtain the site location data corresponding to each candidate location in the target location area.

Step 704: Input the site location data into the site site selection model, and obtain the channel service capability score and the regional development potential score corresponding to each candidate site.

Step 706 , based on the channel service capability score and the regional development potential score, obtain the site selection location corresponding to the target site selection area.

Among them, the site location data can be specifically determined according to the smart features obtained by screening the initial features. After the candidate sites are determined, the site site selection data can be obtained by screening the data corresponding to the candidate sites based on the smart features used to construct the model. .

Specifically, after obtaining the site location model, the site site selection in the target site selection area can be performed based on the site site location model. First, several candidate sites need to be selected from the target site selection area. Then input the site location data corresponding to these candidate locations into the site location model to obtain the channel service capability score and regional development potential score corresponding to each candidate location; based on the channel service capability score and regional development potential score, obtain the target location area corresponding location. In one embodiment, as shown in FIG. 8 , a classification quadrant map based on the channel service capability score and the regional development potential score can be constructed, so as to determine the score quadrant corresponding to each candidate location, so as to determine the corresponding target location area based on the demand. Address location. In another embodiment, the outlet location model of the present application can also be used to evaluate the health of existing outlet channels, thereby optimizing channel distribution, improving resource utilization, and increasing market share. In this embodiment, the site site selection is judged by the site site selection data and site site selection model, and the optimal site site address can be effectively selected from the candidate sites to realize the optimization of the site site site selection process.

It should be understood that, although the steps in the flowcharts involved in the above embodiments are sequentially displayed according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, the execution of these steps is not strictly limited to the order, and these steps may be performed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same time, but may be executed at different times. The order of execution of these steps or stages is also not necessarily sequential, but may be performed alternately or alternately with other steps or at least a portion of the steps or stages in the other steps.

Based on the same inventive concept, an embodiment of the present application further provides an apparatus for constructing a site location model for implementing the above-mentioned method for constructing a site location model. The solution to the problem provided by the device is similar to the implementation described in the above method, so the specific limitations in the embodiment of one or more network site location model building apparatus provided below can refer to the above for the network site location selection. The limitation of the model construction method will not be repeated here.

In one embodiment, as shown in FIG. 9 , an apparatus for constructing a site selection model is provided, including:

The data screening module 902 is configured to obtain the original feature data corresponding to the site selection from the multi-channel multi-source data based on the preset feature index system.

The feature screening module 904 is used to obtain the evidence weight and information value corresponding to each feature in the original feature data, and based on the evidence weight and information value, screen the smart features corresponding to the site location model.

The model building module 906 is configured to build a site selection model by principal component analysis and entropy method based on smart features.

In one embodiment, the data screening module 902 is specifically configured to: obtain the multi-source data of the channel corresponding to the site selection; classify and filter the multi-source data of the channel based on the preset feature index system, and obtain the original feature data corresponding to the site selection.

In one embodiment, the feature screening module 904 is specifically configured to: obtain the null ratios corresponding to various types of feature data in the original feature data; screen the raw feature data based on the null ratios; obtain evidence corresponding to the filtered raw feature data Weight and information value; remove the original feature data whose information value is lower than the preset information value threshold to obtain the first initial feature; identify the nonlinear relationship corresponding to the first initial feature based on the evidence weight, and convert the nonlinear relationship corresponding to the first initial feature The first initial feature of the inverted U-shaped relationship is used as the second initial feature; based on the second initial feature, the smart feature corresponding to the network site location model is obtained.

In one embodiment, the feature screening module 904 is specifically configured to: train an initial random forest model based on a global search method, and obtain a random forest model; and obtain smart features corresponding to the site location model based on the initial features, including: based on the random forest model The initial features are filtered to obtain smart features.

In one embodiment, the model building module 906 is specifically configured to: perform dimension reduction processing on smart features through principal component analysis to obtain dimension reduction features; process the dimension reduction features through entropy method to obtain feature weights corresponding to the dimension reduction features ; Based on the dimensionality reduction features and the feature weights corresponding to the dimensionality reduction features, a network site selection model is constructed.

In one embodiment, the preset feature index system, the channel service dimension index system and the regional development dimension index system, further includes an outlet location module for: acquiring outlet location data corresponding to each candidate location in the target location area; The outlet location data is input into the outlet location model to obtain the channel service capability score and regional development potential score corresponding to each candidate location; based on the channel service capability score and the regional development potential score, the location location corresponding to the target location area is obtained.

All or part of the modules in the above-mentioned network site location model construction device can be implemented by software, hardware and combinations thereof. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

In one embodiment, a computer device is provided, and the computer device may be a server, and its internal structure diagram may be as shown in FIG. 10 . The computer device includes a processor, memory, and a network interface connected by a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes non-volatile storage media and internal memory. The nonvolatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the execution of the operating system and computer programs in the non-volatile storage medium. The database of the computer equipment is used to store the site location model construction data. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for constructing a site selection model is realized.

Those skilled in the art can understand that the structure shown in FIG. 10 is only a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer equipment to which the solution of the present application is applied. Include more or fewer components than shown in the figures, or combine certain components, or have a different arrangement of components.

In one embodiment, a computer device is provided, including a memory and a processor, a computer program is stored in the memory, and the processor implements the following steps when executing the computer program:

Based on the preset feature index system, obtain the original feature data corresponding to the site selection from the multi-source data of the channel;

Obtain the evidence weight and information value corresponding to each feature in the original feature data, and screen the smart features corresponding to the site selection model based on the evidence weight and information value;

Based on smart features, a network site selection model is constructed through principal component analysis and entropy method.

In one embodiment, when the processor executes the computer program, the following steps are further implemented: acquiring channel multi-source data corresponding to the site selection; classifying and screening the channel multi-source data based on a preset feature index system, and acquiring raw data corresponding to the site site selection. characteristic data.

In one embodiment, when the processor executes the computer program, the following steps are further implemented: obtaining the null ratios corresponding to various types of feature data in the original feature data; screening the raw feature data based on the null ratios; obtaining the filtered raw feature data Corresponding evidence weight and information value; remove the original feature data whose information value is lower than the preset information value threshold to obtain the first initial feature; identify the nonlinear relationship corresponding to the first initial feature based on the evidence weight, The first initial feature whose nonlinear relationship is an inverted U-shaped relationship is used as the second initial feature; and based on the second initial feature, the smart feature corresponding to the site selection model is obtained.

In one embodiment, the processor further implements the following steps when executing the computer program: training an initial random forest model based on a global search method to obtain a random forest model; screening initial features based on the random forest model to obtain smart features.

In one embodiment, the processor further implements the following steps when executing the computer program: performing dimension reduction processing on smart features through principal component analysis to obtain dimension reduction features; Based on the feature weights of dimensionality reduction features and the feature weights corresponding to the dimensionality reduction features, a network site selection model is constructed.

In one embodiment, the processor further implements the following steps when executing the computer program: acquiring the site location data corresponding to each candidate location in the target site selection area; Channel service capability score and regional development potential score; based on the channel service capability score and regional development potential score, obtain the location corresponding to the target location area.

In one embodiment, a computer-readable storage medium is provided on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

Based on the preset feature index system, obtain the original feature data corresponding to the site selection from the multi-source data of the channel;

Obtain the evidence weight and information value corresponding to each feature in the original feature data, and screen the smart features corresponding to the site selection model based on the evidence weight and information value;

Based on smart features, a network site selection model is constructed through principal component analysis and entropy method.

In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: acquiring the multi-source data of the channel corresponding to the site selection; raw feature data.

In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: obtaining the null ratios corresponding to various types of feature data in the original feature data; screening the raw feature data based on the null ratios; obtaining the filtered raw features Evidence weight and information value corresponding to the data; remove the original feature data whose information value is lower than the preset information value threshold to obtain the first initial feature; identify the nonlinear relationship corresponding to the first initial feature based on the evidence weight, and assign the first initial feature to the corresponding The first initial feature of the inverted U-shaped relationship is used as the second initial feature; based on the second initial feature, the smart feature corresponding to the site selection model is obtained.

In one embodiment, the computer program further implements the following steps when executed by the processor: training an initial random forest model based on a global search method to obtain a random forest model; screening initial features based on the random forest model to obtain smart features.

In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: performing dimension reduction processing on smart features through principal component analysis to obtain dimension reduction features; processing the dimension reduction features through entropy method to obtain dimension reduction features Corresponding feature weights; based on the dimension reduction features and the feature weights corresponding to the dimension reduction features, a network site selection model is constructed.

In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: acquiring the site location data corresponding to each candidate location in the target site selection area; inputting the site location data into the site location model, and acquiring the corresponding site location data of each candidate location Based on the channel service capability score and regional development potential score; based on the channel service capability score and the regional development potential score, the site selection location corresponding to the target location area is obtained.

In one embodiment, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the following steps:

Based on the preset feature index system, obtain the original feature data corresponding to the site selection from the multi-source data of the channel;

Obtain the evidence weight and information value corresponding to each feature in the original feature data, and screen the smart features corresponding to the site selection model based on the evidence weight and information value;

Based on smart features, a network site selection model is constructed through principal component analysis and entropy method.

In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: acquiring the multi-source data of the channel corresponding to the site selection; raw feature data.

In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: obtaining the null ratios corresponding to various types of feature data in the original feature data; screening the raw feature data based on the null ratios; obtaining the filtered raw features Evidence weight and information value corresponding to the data; identify initial features in original feature data based on evidence weight and information value; obtain smart features corresponding to the site selection model based on initial features.

In one embodiment, the computer program further implements the following steps when executed by the processor: training an initial random forest model based on a global search method to obtain a random forest model; screening initial features based on the random forest model to obtain smart features.

In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: performing dimension reduction processing on smart features through principal component analysis to obtain dimension reduction features; processing the dimension reduction features through entropy method to obtain dimension reduction features Corresponding feature weights; based on the dimension reduction features and the feature weights corresponding to the dimension reduction features, a network site selection model is constructed.

In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: acquiring the site location data corresponding to each candidate location in the target site selection area; inputting the site location data into the site location model, and acquiring the corresponding site location data of each candidate location Based on the channel service capability score and regional development potential score; based on the channel service capability score and the regional development potential score, the site selection location corresponding to the target location area is obtained.

It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all Information and data authorized by the user or fully authorized by the parties.

Those skilled in the art can understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium , when the computer program is executed, it may include the processes of the above-mentioned method embodiments. Wherein, any reference to a memory, a database or other media used in the various embodiments provided in this application may include at least one of a non-volatile memory and a volatile memory. Non-volatile memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetic variable memory (Magnetoresistive Random Memory) Access Memory, MRAM), Ferroelectric Random Access Memory (FRAM), Phase Change Memory (Phase Change Memory, PCM), graphene memory, etc. Volatile memory may include random access memory (Random Access Memory, RAM) or external cache memory, and the like. As an illustration and not a limitation, the RAM can be in various forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM). The database involved in the various embodiments provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a blockchain-based distributed database, etc., but is not limited thereto. The processors involved in the various embodiments provided in this application may be general-purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices, data processing logic devices based on quantum computing, etc., and are not limited to this.

The technical features of the above embodiments can be combined arbitrarily. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described. However, as long as there is no contradiction in the combination of these technical features It is considered to be the range described in this specification.

The above examples only represent several embodiments of the present application, and the descriptions thereof are relatively specific and detailed, but should not be construed as a limitation on the scope of the patent of the present application. It should be noted that, for those skilled in the art, without departing from the concept of the present application, several modifications and improvements can be made, which all belong to the protection scope of the present application. Therefore, the scope of protection of the present application should be determined by the appended claims.

Claims (10)

1. A method for constructing a site selection model of a network point is characterized by comprising the following steps:

acquiring original characteristic data corresponding to website selection from channel multi-source data based on a preset characteristic index system;

acquiring evidence weight and information value corresponding to each feature in the original feature data, and screening smart features corresponding to a site selection model based on the evidence weight and the information value;

and constructing a site selection model of the net points by a principal component analysis and entropy method based on the intelligent characteristics.

2. The method of claim 1, wherein the obtaining of raw feature data corresponding to website addressing from channel multi-source data based on a preset feature index system comprises:

acquiring channel multi-source data corresponding to the site selection of the network points;

and classifying and screening the channel multi-source data based on a preset feature index system to obtain original feature data corresponding to the site selection of the network points.

3. The method according to claim 1, wherein the obtaining of the evidence weight and the information value corresponding to each feature in the raw feature data, and the screening of smart features required by the site selection model based on the evidence weight and the information value comprises:

obtaining null value rates corresponding to various types of feature data in the original feature data;

screening the original characteristic data based on the null value rate;

acquiring evidence weight and information value corresponding to the screened original characteristic data;

removing original characteristic data with information value lower than a preset information value threshold value to obtain a first initial characteristic;

identifying a nonlinear relation corresponding to the first initial feature based on the evidence weight, and taking the first initial feature of which the nonlinear relation corresponding to the first initial feature is an inverse U-shaped relation as a second initial feature;

and acquiring smart characteristics corresponding to the site selection model based on the second initial characteristics.

4. The method according to claim 3, wherein before obtaining the smart feature corresponding to the site selection model based on the second initial feature, further comprising:

training the initial random forest model based on a global search method to obtain a random forest model;

the acquiring of the smart characteristics corresponding to the site selection model based on the second initial characteristics comprises:

and screening the second initial characteristics based on a random forest model to obtain smart characteristics.

5. The method of claim 1, wherein constructing a site selection model by principal component analysis and entropy based on the smart features comprises:

performing dimensionality reduction processing on the smart features through principal component analysis to obtain dimensionality reduction features;

processing the dimensionality reduction features through an entropy method to obtain feature weights corresponding to the dimensionality reduction features;

and constructing a site selection model of the net points based on the dimension reduction features and the feature weights corresponding to the dimension reduction features.

6. The method according to claim 1, wherein the pre-set characteristic index system, the service dimension index system and the region development dimension index system, after constructing the site selection model by principal component analysis and entropy method based on the smart characteristics, further comprises:

acquiring site selection data of network points corresponding to each candidate site in a target site selection area;

inputting the site selection data into the site selection model to obtain channel service capability scores and regional development potential scores corresponding to the candidate sites;

and acquiring the site selection site corresponding to the target site selection area based on the channel service capability score and the area development potential score.

7. A mesh point site selection model construction device is characterized by comprising the following components:

the data screening module is used for acquiring original characteristic data corresponding to the site selection of the network points from channel multi-source data based on a preset characteristic index system;

the characteristic screening module is used for acquiring the evidence weight and the information value corresponding to each characteristic in the original characteristic data and screening the smart characteristic corresponding to the site selection model based on the evidence weight and the information value;

and the model construction module is used for constructing a site selection model of the net points by a principal component analysis and entropy method based on the intelligent characteristics.

8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 6.

9. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 6.

10. A computer program product comprising a computer program, characterized in that the computer program realizes the steps of the method of any one of claims 1 to 6 when executed by a processor.

Images