Data screening method, device, equipment and medium

Abstract

The embodiment of the present application discloses a data screening method, device, equipment and medium. A selection variable is selected from multiple variables to be screened, and the current model is updated according to the selection variable to obtain a target model; when it is detected that the target model does not meet the preset iteration conditions, the total number of variables in the target model is not greater than the total number of preset variables, the second evaluation value of the target model is greater than the first evaluation value of the current model, and the significant level value of each variable in the target model is not greater than the preset significant level value, the target model is determined as the current model, and the selection variable is returned to determine; when the total number is greater than the total number of preset variables, multiple models to be screened are combined according to each variable in the target model, and the model with the highest evaluation value among the multiple models to be screened is used as the target model, and the selection variable is returned to determine; when it is detected that the target model meets the preset iteration conditions, the variable contained in the target model with the highest evaluation value is selected from all target models as the screening variable.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and specifically to a data screening method, device, equipment and medium. Background Art

[0002] With the development of artificial intelligence technology, models can predict corresponding results in some application scenarios. The accuracy of the model's prediction results is related to the input variables. For example, the accuracy of the model's prediction results can be changed by inputting different variables.

[0003] Inputting poor quality variables into the model will cause the accuracy of the model's prediction results to deteriorate. Inputting other variables of better quality into the model will cause the accuracy of the model's prediction results to improve. Therefore, it is necessary to reasonably screen the variables input into the model so that the model's prediction results can maintain a high degree of accuracy.

[0004] In the related art, in the variable screening stage corresponding to the model, although adding variables or deleting variables is adopted to improve the accuracy of the model's prediction results, variable screening is often implemented based on manual experience or one-by-one traversal, resulting in an inability to efficiently determine the variables that need to be added or deleted in the model. Therefore, the efficiency of variable screening in the related art is low. Summary of the invention

[0005] The embodiments of the present application provide a data screening method, apparatus, device and medium, which can improve the efficiency of screening data input into a model.

[0006] In order to achieve the above-mentioned purpose, according to one aspect of the present application, an embodiment of the present application provides a data screening method, comprising:

[0007] Acquire multiple variables to be screened, select a selection variable from the multiple variables to be screened, and update the current model according to the selection variable to obtain a target model;

[0008] When it is detected that the target model does not satisfy a preset iteration condition, determining a first evaluation value corresponding to the current model and a second evaluation value corresponding to the target model;

[0009] When the total number of variables in the target model is not greater than the preset total number of variables, the second evaluation value is greater than the first evaluation value, and the significance level value of each variable in the target model is not greater than the preset significance level value, the target model is determined as the current model, and the process returns to determine the selected variable from the multiple variables to be screened;

[0010] When the total number is greater than the total number of preset variables, a plurality of models to be screened are combined according to each variable in the target model, and the model with the highest evaluation value among the plurality of models to be screened is used as the updated target model, the updated target model is determined as the current model, and the execution is returned to select a selection variable from the plurality of variables to be screened;

[0011] When it is detected that the target model meets the preset iteration condition, the target model with the highest evaluation value is selected from the historical target models, and the variables included in the target model with the highest evaluation value are determined as screening variables.

[0012] In order to achieve the above-mentioned purpose, according to one aspect of the present application, an embodiment of the present application provides a data screening device, including:

[0013] An acquisition module is used to acquire multiple variables to be screened, select a selection variable from the multiple variables to be screened, and update the current model according to the selection variable to obtain a target model;

[0014] A first determination module, configured to determine a first evaluation value corresponding to the current model and a second evaluation value corresponding to the target model when it is detected that the target model does not satisfy a preset iteration condition;

[0015] A second determination module is used to determine the target model as the current model and return to execute the determination of the selected variable from the multiple variables to be screened when the total number of variables in the target model is not greater than the total number of preset variables, the second evaluation value is greater than the first evaluation value, and the significance level value of each variable in the target model is not greater than the preset significance level value;

[0016] A third determination module is used for, when the total number is greater than the total number of preset variables, combining multiple models to be screened according to each variable in the target model, and taking the model with the highest evaluation value among the multiple models to be screened as the updated target model, determining the updated target model as the current model, and returning to execute selecting a selection variable from the multiple variables to be screened;

[0017] The screening module is used to select the target model with the highest evaluation value from the historical target models when it is detected that the target model meets the preset iteration condition, and determine the variables contained in the target model with the highest evaluation value as the screening variables.

[0018] In some embodiments, the third determination module includes a to-be-deleted submodule, a selection submodule, and a combination submodule;

[0019] A to-be-deleted submodule is used for, when it is detected that there is a variable with a significance level value greater than the preset significance level value in the target model, determining the variable with a significance level value greater than the preset significance level value in the target model as a to-be-deleted variable;

[0020] A selection submodule, for determining the variables in the target model whose significance level value is not greater than the preset significance level value as selected variables;

[0021] The combination submodule is used to combine multiple groups of input variables according to the variables to be deleted and the selected variables, and input each group of input variables into the basic model to obtain multiple models to be screened, wherein the basic model is a model without variables.

[0022] In some embodiments, the combination submodule is further configured to:

[0023] When there is no variable in the target model whose significance level value is greater than the preset significance level value, combining multiple groups of input variables according to each variable in the target model;

[0024] Each set of input variables is input into the basic model to obtain multiple models to be screened.

[0025] In some embodiments, the screening module is used to:

[0026] Determine a first target model from the historically obtained target models, the total number of variables of which is not greater than the total number of the preset variables;

[0027] Determine a second target model in which the significance level value of each variable is not greater than the preset significance level value in the first target model;

[0028] A target model with the highest evaluation value is determined in the second target model, and a variable of the target model with the highest evaluation value is determined as a screening variable.

[0029] In some embodiments, the screening module is further used to:

[0030] After obtaining the target model each time, obtaining the previous target model;

[0031] When the variables contained in the previous target model are the same as the variables contained in the target model, it is determined that the target model meets the preset iteration conditions, and the target model with the highest evaluation value is selected from the historical target models, and the variables contained in the target model with the highest evaluation value are determined as screening variables.

[0032] In some implementations, the second determining module is further configured to:

[0033] When it is detected that the total number of variables in the target model is not greater than the preset total number of variables, if the second evaluation value is not greater than the first evaluation value, or the significance level value of any variable in the target model is greater than the preset significance level value, then returning to execute selecting a selected variable from the multiple variables to be screened;

[0034] The current model is updated according to the selected variables to obtain a target model.

[0035] In some embodiments, the second determination module further includes a detection submodule and a determination submodule;

[0036] A detection submodule, used for returning to execute selecting a selection variable from the plurality of variables to be screened when detecting that there is a model identical to the target model in the historical target models before determining the target model as the current model;

[0037] The determination submodule is used to determine the target model as the current model when it is determined that there is no model identical to the target model in the target models obtained in the history.

[0038] In some embodiments, the acquisition module includes an arrangement submodule and an acquisition submodule;

[0039] An arrangement submodule, used for determining a significance level value corresponding to each variable to be screened among the multiple variables to be screened before selecting a selection variable from the multiple variables to be screened;

[0040] Arrange each of the variables to be screened from low to high according to the significance level value to obtain a sequence of variables to be screened;

[0041] The multiple variables to be screened are processed in batches according to the sequence of variables to be screened to obtain multiple batches of variable sets to be screened.

[0042] Get the submodule to determine the set of variables to be screened corresponding to the current batch;

[0043] When there are unselected variables to be screened in the set of variables to be screened corresponding to the current batch, a selected variable is selected from the unselected variables to be screened;

[0044] When there are no unselected variables to be screened in the set of variables to be screened corresponding to the current batch, a selected variable is selected from the set of variables to be screened corresponding to the next batch.

[0045] In order to achieve the above-mentioned purpose, according to one aspect of the present application, an embodiment of the present application provides a computer-readable storage medium, which stores multiple instructions, and the instructions are suitable for a processor to load to execute the data screening method provided by the embodiment of the present application.

[0046] In order to achieve the above-mentioned purpose, according to one aspect of the present application, an embodiment of the present application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the data screening method provided in the embodiment of the present application when executing the computer program.

[0047] In order to achieve the above-mentioned purpose, according to one aspect of the present application, an embodiment of the present application provides a computer program product, including a computer program or instructions, which, when executed by a processor, implements the data screening method provided by the embodiment of the present application.

[0048] In an embodiment of the present application, a plurality of variables to be screened are obtained, a selection variable is selected from the plurality of variables to be screened, and the current model is updated according to the selection variable to obtain a target model; when it is detected that the target model does not satisfy a preset iteration condition, a first evaluation value corresponding to the current model and a second evaluation value corresponding to the target model are determined; when the total number of variables in the target model is not greater than the total number of preset variables, the second evaluation value is greater than the first evaluation value, and the significance level value of each variable in the target model is not greater than the preset significance level value, the target model is determined as the current model, and the process of determining the selection variable from the plurality of variables to be screened is returned to execute; when the total number is greater than the total number of preset variables, a plurality of models to be screened are combined according to each variable in the target model, and the model with the highest evaluation value from the plurality of models to be screened is used as the updated target model, the updated target model is determined as the current model, and the process of selecting the selection variable from the plurality of variables to be screened is returned to execute; when it is detected that the target model satisfies the preset iteration condition, a target model with the highest evaluation value is selected from the target models obtained historically, and the variables contained in the target model with the highest evaluation value are determined as screening variables.

[0049] Thus, by setting the total number of preset variables of the variables of the model to be obtained, when the total number of variables of the target model is not greater than the total number of preset variables, the selected variables can be determined from multiple variables to be screened, and the target model is updated by the selected variables, so as to obtain a target model with a higher evaluation value than the previous model, so as to efficiently increase the variables in the target model; on the contrary, when the total number of variables in the target model is greater than the total number of preset variables, multiple models to be screened are combined according to each variable in the target model, and the model with the highest evaluation value among the multiple models to be screened is used as the updated target model, so as to efficiently reduce the redundant variables in the target model, obtain the updated target model, and then determine from the multiple variables to be screened. Determine the selection variables, and then update the target model by selecting the variables, so as to realize the continuous updating of the target model. In the process of updating the target model by adding or deleting variables, the target model with better performance than the current model can be continuously determined according to the evaluation value of the target model and the significance level value of each variable in the target model. Finally, when the target model meets the preset iteration conditions, the target model with the highest evaluation value can be determined from multiple target models obtained in history, and the variables contained in the target model with the highest evaluation value are used as the final screening variables. Compared with the related technology of realizing variable screening based on manual experience or traversal one by one, the efficiency of variable screening in this application is higher, thereby improving the data screening efficiency.

[0050] Other features and advantages of the present application will be described in the following description, and partly become apparent from the description, or understood by practicing the present application. The purpose and other advantages of the present application can be realized and obtained by the structures specifically pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0052] Figure 1 is a schematic diagram of a system framework corresponding to the data screening method provided in an embodiment of the present application;

[0053] Figure 2 is a schematic diagram of a scenario corresponding to the data screening method provided in an embodiment of the present application;

[0054] Figure 3 It is a flow chart of the data screening method provided in the embodiment of the present application;

[0055] Figure 4It is a schematic diagram of a process for determining a model to be screened provided in an embodiment of the present application;

[0056] Figure 5 is a schematic diagram of a process for determining screening variables provided in an embodiment of the present application;

[0057] Figure 6 is another flow chart of the data screening method provided in an embodiment of the present application;

[0058] Figure 7 is another flow chart of the data screening method provided in an embodiment of the present application;

[0059] Figure 8 is a structural schematic diagram of a data screening device provided in an embodiment of the present application;

[0060] Fig. 9 It is a schematic diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0061] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0062] It should be noted that in each specific implementation of the present application, when it comes to the need to process data related to the identity or characteristics of an object based on variables, the permission or consent of the object will be obtained first, and the collection, use and processing of these data will comply with relevant laws, regulations and standards. In addition, when the embodiment of the present application needs to obtain variables, the separate permission or separate consent of the object will be obtained through a pop-up window or jump to a confirmation page. After clearly obtaining the separate permission or separate consent of the object, the necessary object-related data for enabling the normal operation of the embodiment of the present application is obtained.

[0063] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0064] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0065] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0066] Before further describing the embodiments of the present application in detail, the nouns and terms involved in the embodiments of the present application are explained. The nouns and terms involved in the embodiments of the present application are subject to the following interpretations:

[0067] Significance level value: ‌also known as P-value, is a concept in statistics, ‌used to measure whether the association between observed data and a hypothesis is statistically significant. ‌It indicates the probability of the observed statistic or more extreme situation occurring when the null hypothesis is true. ‌When the P-value is very small, it means that the probability of obtaining the current sample result is very low when the null hypothesis is true. This means that the null hypothesis is unlikely to be true, and thus tends to reject the null hypothesis. When the P-value is large, there is not enough evidence to reject the null hypothesis. ‌In this application, the smaller the P-value of a variable, the more significant the effect of the variable on the prediction results of the model, and the larger the P-value of a variable, the less significant the effect of the variable on the prediction results of the model. In other words, the smaller the P-value of a variable, the greater its positive impact on the prediction results of the model, which can improve the accuracy of the prediction results of the model.

[0068] The evaluation value of the model: KS value (Kolmogorov-Smirnov value) is an indicator to measure the risk differentiation ability of the model. It measures the difference between the cumulative distribution of good and bad samples. In statistics and data analysis, KS value (Kolmogorov-Smirnov value) is also called Kolmogorov-Smirnov test value. In the financial field, KS value is often used to evaluate the accuracy of the model in predicting the probability of default. The larger the KS value, the greater the difference in the cumulative distribution of good and bad samples, and the stronger the risk differentiation ability of the model. By calculating the KS value, we can find the optimal threshold, thereby optimizing the prediction effect of the model.

[0069] For example, in credit evaluation, by comparing the KS values ​​of the actual default distribution and the default distribution predicted by the model, we can judge the model's ability to distinguish between defaulting customers and non-defaulting customers. If the KS value is high, it means that the model can better distinguish between the two types of customers; if the KS value is low, it means that the model's ability to distinguish is weak.

[0070] Logistic regression model: The logistic regression model maps the input feature variables to a probability value by establishing a logistic function, which represents the possibility of belonging to a specific category. It is based on the logit function, that is, for a binary classification problem, assuming that the probability of a sample belonging to the positive class is P, then the probability of belonging to the negative class is 1-P.

[0071] First, let’s describe the technical problems existing in the relevant technologies:

[0072] With the development of artificial intelligence technology, models can predict corresponding results in some application scenarios. The accuracy of the model's prediction results is related to the input variables. For example, the accuracy of the model's prediction results can be changed by inputting different variables.

[0073] Inputting poor quality variables into the model will cause the accuracy of the model's prediction results to deteriorate. Inputting other variables of better quality into the model will cause the accuracy of the model's prediction results to improve. Therefore, it is necessary to reasonably screen the variables input into the model so that the model's prediction results can maintain a high degree of accuracy.

[0074] In the related art, in the variable screening stage corresponding to the model, although adding variables or deleting variables is adopted to improve the accuracy of the model's prediction results, variable screening is often implemented based on manual experience or one-by-one traversal, resulting in an inability to efficiently determine the variables that need to be added or deleted in the model. Therefore, the efficiency of variable screening in the related art is low.

[0075] In order to solve the above technical problems, in an embodiment of the present application, by setting a preset total number of variables of the model that needs to be obtained, when the total number of variables of the target model is not greater than the preset total number of variables, a selection variable can be determined from multiple variables to be screened, and the target model is updated by the selection variable, so as to obtain a target model with a higher evaluation value than the previous model, so as to efficiently increase the variables in the target model; on the contrary, when the total number of variables in the target model is greater than the preset total number of variables, multiple models to be screened are combined according to each variable in the target model, and the model with the highest evaluation value among the multiple models to be screened is used as the updated target model, so as to efficiently reduce the redundant variables in the target model, obtain the updated target model, and then The selected variables are determined from the variables to be screened, and the target model is then updated by the selected variables, so as to realize continuous updating of the target model. In the process of updating the target model by adding or deleting variables, the target model with better performance than the current model can be continuously determined according to the evaluation value of the target model and the significance level value of each variable in the target model. Finally, when the target model meets the preset iteration conditions, the target model with the highest evaluation value can be determined from multiple target models obtained in history, and the variables contained in the target model with the highest evaluation value are used as the final screening variables. Compared with the related art of realizing variable screening based on manual experience or traversal one by one, the variable screening in this application is more efficient, thereby improving the data screening efficiency.

[0076] Specifically, the embodiments of the present application provide a data screening method, device, equipment and medium. Specifically, the embodiments of the present application will be described from the dimension of a data screening device, which can be integrated in a computer device, which can be a server or a terminal and other devices. Among them, the server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. Among them, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc., but is not limited to this. The embodiments of the present application can be applied to various scenarios, including but not limited to cloud computing, cloud services, data screening, financial services and other scenarios.

[0077] See also Figure 1 , Figure 1 1 is a schematic diagram of a system framework corresponding to the data screening method provided in the embodiment of the present application. The data screening method provided in the embodiment of the present application can be applied to the system framework.

[0078] Please refer to Figure 1 , Figure 1 1 is a system architecture diagram used by the data screening method provided in the embodiment of the present application, which includes a terminal 140, the Internet 130, a gateway 120, a server 110, and the like.

[0079] The terminal 140 or the server 110 may be a device for executing the data screening method.

[0080] The terminal 140 includes but is not limited to mobile phones, computers, intelligent voice interaction devices, smart home appliances, vehicle-mounted terminals, aircraft, etc. The embodiments of the present application can be applied to various scenarios, including but not limited to cloud office, enterprise management, etc. In addition, it can be a single device or a collection of multiple devices. For example, multiple desktop computers are interconnected through a local area network, share a display, etc. to work together, and together constitute a terminal 140. The terminal 140 can communicate with the Internet 130 in a wired or wireless manner to exchange data.

[0081] The server 110 refers to a computer system that can provide certain services to the terminal 140. Compared with the ordinary terminal 140, the server 110 has higher requirements in terms of stability, security, performance, etc. The server 110 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms.

[0082] The gateway 120 is also called an internetwork connector or a protocol converter. The gateway realizes network interconnection at the transport layer and is a computer system or device that acts as a converter. The gateway is a translator between two systems that use different communication protocols, data formats or languages, or even completely different architectures. At the same time, the gateway can also provide filtering and security functions. The message sent by the terminal 140 to the server 110 must be sent to the corresponding server 110 through the gateway 120. The message sent by the server 110 to the terminal 140 must also be sent to the corresponding terminal 140 through the gateway 120.

[0083] The data screening method in the embodiment of the present application can be applied in a variety of scenarios, such as cloud computing, cloud services, data screening, financial services, etc. The scenarios to which the data screening method in the present application is applied are not limited here.

[0084] See also Figure 2 , Figure 2 It is a schematic diagram of a scenario corresponding to the data screening method provided in an embodiment of the present application.

[0085] The variables input into the model directly affect the quality of the model's prediction results. Therefore, it is necessary to determine the best combination of variables to be input into the model from multiple screening variables, so that the model's prediction results are better or the model's prediction results meet actual needs.

[0086] Among the multiple screening variables, there are variable 1, variable 2, variable 3...variable N, each variable can be a variable of a data type, for example, the data type of variable 1 is age, the data type of variable 2 is education level, and the data type of variable 3 is region. Therefore, among the multiple screening variables, it can be considered that the data type of each variable is different from the data type of other variables.

[0087] After obtaining multiple screening variables, a selection variable can be determined from the screening variables. For example, variable 1 and variable 3 can be selected as selection variables, or variable 2 and variable 4 can be selected as selection variables. The selection variable is at least one variable selected from multiple screening variables.

[0088] After the selected variables are determined, the selected variables can be input into the current model, and then the current model can be updated to obtain the target model. The current model can be a basic model, that is, a model without input variables, such as a logistic regression model or a linear regression model. The current model can also be a model that includes selected variables, such as the selected variables have been input into the basic model.

[0089] After obtaining the target model, if the total number of variables in the target model is not greater than the total number of preset variables, determine whether the target model is a target model with better performance than the current model based on the evaluation value (ks value) of the target model and the significance level value (p value) of each variable in the target model. If the target model is a target model with better performance than the current model, determine the target model as the current model, and then continue to determine the selection variables among multiple variables to be screened, and then update the current model based on the selection variables to obtain the updated target model.

[0090] After obtaining the target model, if the total number of variables in the target model is greater than the total number of preset variables, some variables in the target model need to be deleted to update the target model and obtain an updated target model. The total number of preset variables is the maximum total number of variables that can be input in the model, and the total number of preset variables can be set according to the actual needs of the model.

[0091] Among them, the above content briefly describes the updating process of the target model. In this application, the total number of variables in each updated target model is less than the total number of preset variables, and the target model with better performance can be accurately screened out through the evaluation value of the target model and the significant level value of the variable. For example, the target model with the highest evaluation value is determined among target model 1, target model 2...target model M. Finally, among the multiple target models obtained historically, the target model with the highest evaluation value can be determined, and the variables contained in the target model with the highest evaluation value are determined as the screened screening variables.

[0092] The model in this application can be a scoring card model, which can be specifically applied in financial service scenarios. For example, for different objects, some characteristics of the objects can be obtained as variables, such as region, age, marital status, education level, and monthly debt ratio, and then input into the model to output the corresponding score value of the object. Finally, it can be determined whether to provide corresponding financial services to the object based on the score value.

[0093] The basic model corresponding to the model in the present application can be a logistic regression model or a linear regression model, without any specific limitation, and the scenario in which the model is applied can be set according to actual needs without any limitation.

[0094] The data screening method, device, equipment and medium provided by this application will be described in detail below.

[0095] See also Figure 3 , Figure 3 : is a flow chart of a data screening method provided in an embodiment of the present application. The data screening method may include the following steps:

[0096] Step 210: obtaining multiple variables to be screened, selecting a selection variable from the multiple variables to be screened, and updating the current model according to the selection variable to obtain a target model;

[0097] Step 220: When it is detected that the target model does not meet the preset iteration condition, determine the first evaluation value corresponding to the current model and the second evaluation value corresponding to the target model;

[0098] Step 230: when the total number of variables in the target model is not greater than the preset total number of variables, the second evaluation value is greater than the first evaluation value, and the significance level value of each variable in the target model is not greater than the preset significance level value, the target model is determined as the current model, and the process returns to determine the selected variable from the multiple variables to be screened;

[0099] Step 240: when the total number is greater than the total number of preset variables, multiple models to be screened are combined according to each variable in the target model, and the model with the highest evaluation value among the multiple models to be screened is used as the updated target model, and the updated target model is determined as the current model, and the process of selecting a selected variable from the multiple variables to be screened is returned to the execution;

[0100] Step 250: When it is detected that the target model meets the preset iteration condition, the target model with the highest evaluation value is selected from the historical target models, and the variables included in the target model with the highest evaluation value are determined as screening variables.

[0101] Steps 210 to 250 will be described in detail below.

[0102] In step 210, a plurality of variables to be screened are obtained, a selection variable is selected from the plurality of variables to be screened, and the current model is updated according to the selection variable to obtain a target model.

[0103] In the present application, multiple variables to be screened can be obtained, and each variable to be screened can specifically correspond to a variable of a data type, for example, the data type corresponding to variable to be screened 1 is age, the data type corresponding to variable to be screened 2 is gender, the data type corresponding to variable to be screened 3 is region, the data type corresponding to variable to be screened N is education level, etc., and then the selection variables are determined among these variables to be screened, for example, variables to be screened 1 and variables to be screened 2 are determined as selection variables, and then the selection variables are input into the current model, thereby updating the current model and obtaining the target model.

[0104] In some implementations, before selecting a selection variable from a plurality of variables to be screened, the method further includes:

[0105] (1.1) Determine the significance level value corresponding to each variable among multiple variables to be screened;

[0106] (1.2) Arrange each variable to be screened from low to high according to the significance level value to obtain a sequence of variables to be screened;

[0107] (1.3) The multiple variables to be screened are processed in batches according to the sequence of the variables to be screened, so as to obtain multiple batches of variable sets to be screened.

[0108] Among them, among multiple variables to be screened, all the variables to be screened can be input into the basic model to fit a test model, and the significance level value of each variable, that is, the P-value value of each variable, can be obtained after the hypothesis test is performed on the test model. In the model provided in this application, the significance level value can be used to test the significance of each coefficient in the model. If the significance level value corresponding to a certain variable is very small, it means that the variable has a significant effect on the dependent variable; otherwise, it means that the variable has an insignificant effect on the dependent variable and can be considered to be removed from the model.

[0109] Then, each variable to be screened is arranged from low to high according to the significance level value to obtain a sequence of variables to be screened. For example, the sequence of variables to be screened is [X1, X2, X3, X4, X5...Xn], where the significance level value of variable X1 is smaller than the significance level value of variable X2, the significance level value of variable X2 is smaller than the significance level value of variable X3, and the significance level value of variable Xn-1 is smaller than the significance level value of variable Xn.

[0110] Then, multiple variables to be screened are processed in batches according to the sequence of variables to be screened. For example, variables X1, X2, and X3 are one batch, and variables X4, X5, and X6 are another batch. The batches can be divided according to actual needs to obtain multiple batches of variable sets to be screened.

[0111] In this way, when selecting variables from multiple variables to be screened, the selected variables can be selected from each batch from the front to the back according to the order of the batches. This allows variables with lower significance levels to be selected first, thereby quickly updating some target models with higher evaluation values ​​(ks values), and then selecting variables with higher significance levels, and updating the current model with these variables, thereby obtaining a target model with a higher evaluation value.

[0112] It can be seen from the above that by sorting and batch processing the variables to be screened in the above manner, the efficiency of determining the target model with a higher evaluation value can be improved.

[0113] In some embodiments, selecting a selection variable from a plurality of variables to be screened includes:

[0114] (2.1) Determine the set of variables to be screened corresponding to the current batch;

[0115] (2.2) When there are unselected variables to be screened in the set of variables to be screened corresponding to the current batch, a selected variable is selected from the unselected variables to be screened;

[0116] (2.3) When there are no unselected variables to be screened in the set of variables to be screened corresponding to the current batch, a selected variable is selected from the set of variables to be screened corresponding to the next batch.

[0117] Among the multiple variables to be screened, the set of variables to be screened corresponding to the current batch can be determined first, and then it can be determined whether there are unselected variables to be screened in the set of variables to be screened corresponding to the current batch. If there are unselected variables to be screened in the set of variables to be screened corresponding to the current batch, the selected variable is selected from the unselected variables to be screened. For example, the set of variables to be screened in the current batch is [X4, X5, X6], and variable X6 is not selected, then variable X6 is used as the selected variable.

[0118] If there are no unselected variables to be screened in the set of variables to be screened corresponding to the current batch, then the selected variables are selected from the set of variables to be screened corresponding to the next batch. For example, if the set of variables to be screened in the next batch is [X7, X8, X9], then the selected variables can be determined from the variables X7, X8, and X9.

[0119] In this way, the variables to be screened in each batch can be selected and used as selection variables, and then the selection variables are input into the current model to update the current model and obtain the target model. This can avoid the omission of the variables to be screened among multiple variables to be screened, and can improve the efficiency of determining the selection variables, thereby improving the efficiency of data screening.

[0120] In step 220, when it is detected that the target model does not satisfy the preset iteration condition, a first evaluation value corresponding to the current model and a second evaluation value corresponding to the target model are determined.

[0121] In the present application, corresponding iteration conditions are set for the target model in advance. For example, if the preset iteration condition is that the target model is updated a preset number of times, it is considered that the preset iteration condition is met, and data screening is stopped at this time. For example, after the target model is updated, the variables in the target model for this update are determined, and then the variables in the previous target model are determined. If the variables in the target model for this update are the same as the variables in the previous target model, then it is considered that the target model for this update has met the preset iteration condition, and data screening is stopped at this time. In the preset iteration conditions in these two examples, as long as the target model reaches any preset iteration condition during the data screening process, data screening is stopped.

[0122] When it is detected that the target model does not meet the preset iteration conditions, the first evaluation value corresponding to the current model and the second evaluation value corresponding to the target model are determined. For example, if there are multiple preset iteration conditions and none of these preset iteration conditions are met, it is necessary to continue to determine the selection variable from the multiple variables to be screened, so as to obtain a new target model by selecting the variable.

[0123] Among them, the first evaluation value corresponding to the current model and the second evaluation value corresponding to the target model can be determined first. The first evaluation value and the second evaluation value can be specifically KS values. In data analysis and modeling, KS (Kolmogorov-Smirnov) values ​​are often used to evaluate the distinguishing ability of the model. Generally speaking, the larger the KS value of the model, the better the distinguishing ability of the model. Therefore, it is necessary to first determine the first evaluation value corresponding to the current model and the second evaluation value corresponding to the target model, and use the first evaluation value and the second evaluation value to evaluate whether the performance of the target model is better or worse than that of the current model. The current model obtains the model of the target model by selecting variables to update.

[0124] In some implementations, after comparing the first evaluation value corresponding to the current model with the second evaluation value corresponding to the target model, the following steps may be included:

[0125] (1.1) When it is detected that the total number of variables in the target model is not greater than the preset total number of variables, if the second evaluation value is not greater than the first evaluation value, or the significance level value of any variable in the target model is greater than the preset significance level value, then return to execute the selection of the selected variable from the multiple variables to be screened;

[0126] (1.2) Update the current model according to the selected variables to obtain the target model.

[0127] Among them, firstly, it is detected whether the total number of variables in the target model is greater than the total number of preset variables. When it is detected that the total number of variables in the target model is not greater than the total number of preset variables, it means that the selected variables can be further determined in the variables to be screened, and then the target model is updated by selecting the variables. The total number of preset variables is the total number of variables that can be input in the model at most, and the total number of preset variables can be set according to the actual needs of the model.

[0128] Specifically, it is determined whether the second evaluation value corresponding to the target model is greater than the first evaluation value corresponding to the current model. If the second evaluation value corresponding to the target model is not greater than the first evaluation value corresponding to the current model, that is, the second evaluation value is less than or equal to the first evaluation value, then it means that the target model has a worse performance than the current model, and the target model cannot be used as the basis for the next update of the target model. At this time, the current model can still be used as the basis, and then the selection variable is selected from the multiple variables to be screened, and the current model is updated according to the selection variable to obtain the target model.

[0129] In the present application, the smaller the P value of a variable, the more significant the influence of the variable on the prediction results of the model is considered to be, and the larger the P value of the variable, the less significant the influence of the variable on the prediction results of the model is considered to be. In other words, the smaller the P value of the variable, the greater its positive influence on the prediction results of the model, which can improve the accuracy of the prediction results of the model.

[0130] Therefore, in this application, a preset significance level value is set, and the preset significance level value is a critical standard for measuring the strength of the positive impact of the variable on the prediction results of the model. When the significance level value of the variable is not greater than the preset significance level value, it is considered that the variable has a strong positive impact on the prediction results of the model, that is, the prediction accuracy of the model can be improved. When the significance level value of the variable is greater than the preset significance level value, it is considered that the variable has a weak positive impact on the prediction results of the model, that is, the impact on the prediction accuracy of the model is relatively weak.

[0131] When it is detected that the total number of variables in the target model is not greater than the total number of preset variables, and the significance level value of any variable in the target model is greater than the preset significance level value, since a variable with a larger significance level value is introduced, the variable has a lower impact on the accuracy of the prediction results of the target model. The target model is not a better model than the current model, and the target model cannot be used as the basis for the next model update. At this time, return to select the selection variable from multiple variables to be screened, and update the current model according to the selection variable to obtain the target model.

[0132] The advantage of doing this is that the next updated target model can be updated based on a better model, which can improve the updating efficiency of the target model, that is, the updated target model is always updated in a better direction, making the prediction ability of the updated target model more accurate. In this way, when the subsequent target model meets the preset iteration conditions, the target model with the highest evaluation value can be determined from the multiple target models obtained in history, and the variables contained in the target model with the highest evaluation value can be determined as screening variables, so that the screening of multiple variables to be screened can be achieved efficiently.

[0133] In step 230, when the total number of variables in the target model is not greater than the preset total number of variables, the second evaluation value is greater than the first evaluation value, and the significance level value of each variable in the target model is not greater than the preset significance level value, the target model is determined as the current model, and the execution is returned to determine the selected variable from multiple variables to be screened.

[0134] Among them, when the total number of variables in the target model is not greater than the total number of preset variables, it means that at this time, you can continue to determine the selection variables from the variables to be screened, and then update the target model by selecting the variables.

[0135] If the second evaluation value is greater than the first evaluation value, and the significance level value of each variable in the target model is not greater than the preset significance level value, it means that the target model has a higher evaluation value than the current model, and the target model is better than the current model. At this time, the target model is determined as the current model, and then the execution returns to select the selection variable from multiple variables to be screened, and the current model is updated according to the selection variable to obtain the target model, which is the updated target model.

[0136] For example, if the second evaluation value of the target model is 500, and the first evaluation value of the current model is 350, then the second evaluation value is greater than the first evaluation value, and the target model has better performance than the current model, for example, it has better discrimination ability. At this time, the target model is updated to the current model, and the variables in the current model are determined to be retained variables. Then, the selection variables are selected from multiple variables to be screened, and the current model is updated according to the selection variables to obtain the updated target model. In this way, the target model is updated.

[0137] Then, the second evaluation value of the updated target model and the first evaluation value of the current model are obtained, and then the process returns to execute the detection of whether the total number of variables in the target model is greater than the preset total number of variables. If the total number of variables in the target model is not greater than the preset total number of variables, it is determined whether the second evaluation value corresponding to the target model is greater than the first evaluation value corresponding to the current model. If the second evaluation value corresponding to the target model is not greater than the first evaluation value corresponding to the current model, the next updated target model is obtained in the above manner.

[0138] The advantage of doing this is that the target model to be updated next time can be updated based on the model with a higher evaluation value in the previous update, which can improve the updating efficiency of the target model, that is, the updated target model is always updated in the direction of a higher evaluation value, so that when the subsequent target model meets the preset iteration conditions, the target model with the highest evaluation value can be determined from the multiple target models obtained historically, and the variables contained in the target model with the highest evaluation value can be determined as screening variables, so that the screening of multiple variables to be screened can be achieved efficiently.

[0139] In some embodiments, if the total number of variables in the target model is not greater than the total number of preset variables, and there are variables in the target model whose significance level values ​​are greater than the preset significance level values, then the target model cannot be used as the basis for the next model update, because variables with significance level values ​​higher than the preset significance level values ​​are added, and the significance level values ​​of these variables are high, which actually have little impact on the prediction results of the target model and the generation of the prediction results, so the target model containing the variables cannot be used as the basis for the next model update. At this time, the current model is still used as the basis for the next model update, and then the selection variables are selected from multiple variables to be screened, and the current model is updated according to the selection variables to obtain the target model, thereby updating the target model.

[0140] Similarly, after obtaining the updated target model, first determine whether the total number of variables in the target model is greater than the preset total number of variables. If the total number of variables in the target model is not greater than the preset total number of variables, then determine whether there are variables in the target model whose significant level values ​​are greater than the preset significant level values. If there are variables in the target model whose significant level values ​​are greater than the preset significant level values, then obtain the next updated target model in the above manner.

[0141] In some embodiments, if the total number of variables in the target model is not greater than the total number of preset variables, if the second evaluation value corresponding to the target model is not greater than the first evaluation value corresponding to the current model, and there are variables in the target model whose significance level values ​​are greater than the preset significance level values, then this target model cannot be used as the basis for the next model update, and the current model is still used as the basis for the next model update. Then, the selection variable is selected from multiple variables to be screened, and the current model is updated according to the selection variable to obtain the target model, thereby updating the target model.

[0142] In some embodiments, if the total number of variables in the target model is not greater than the total number of preset variables, if the second evaluation value corresponding to the target model is greater than the first evaluation value corresponding to the current model, but there are variables in the target model whose significance level values ​​are greater than the preset significance level values, then this target model cannot be used as the basis for the next model update, and the current model is still used as the basis for the next model update. Then, the selection variable is selected from multiple variables to be screened, and the current model is updated according to the selection variable to obtain the target model, thereby updating the target model.

[0143] The advantage of doing this is that the significance level value of each variable in the updated target model can be made smaller than the preset significance level value. In this way, in the process of adding selected variables, the updated target model only includes variables that are more relevant to the prediction results of the target model, and a target model with a higher evaluation value can be determined, that is, a target model with more accurate prediction results, such as a more accurate prediction result.

[0144] In some implementations, before determining the target model as the current model, when it is detected that there is a model identical to the target model in the target models obtained historically, the process returns to select a selection variable from a plurality of variables to be screened. When it is detected that there is a model identical to the target model in the target models obtained historically, it indicates that the target model has been generated. In this case, the target model is not determined as the current model, the process returns to select a selection variable from a plurality of variables to be screened, and the current model is updated according to the selection variable to obtain the target model.

[0145] When it is determined that there is no model identical to the target model in the historical target models, the target model is determined as the current model, and then the execution is returned to select a selection variable from multiple variables to be screened, and the current model is updated according to the selection variable to obtain the target model.

[0146] In step 240, when the total number is greater than the total number of preset variables, multiple models to be screened are combined according to each variable in the target model, and the model with the highest evaluation value among the multiple models to be screened is used as the updated target model, and the updated target model is determined as the current model, and the execution is returned to select the selected variable from the multiple variables to be screened.

[0147] It can be understood that in the embodiment of the present application, if the updated target model is obtained by adding and selecting variables, the total number of variables in the updated target model will also increase, and the total number of variables in the target model will exceed the total number of preset variables. In this case, it is necessary to delete some of the variables in the target model to continue updating the target model.

[0148] Specifically, multiple models to be screened can be combined according to each variable in the target model, and the model with the highest evaluation value among the multiple models to be screened is used as the updated target model, and then the variables in the target model are determined as screening variables.

[0149] Among them, when the total number of variables in the target model is greater than the total number of preset variables, it means that the total number of variables entering the target model at this time is greater than the maximum total number of variables entering the model. At this time, the redundant variables in the target model need to be deleted to obtain an updated target model.

[0150] Please also read Figure 4 , Figure 4 is a flow chart of determining a model to be screened provided in an embodiment of the present application. In some embodiments, combining multiple models to be screened according to each variable in the target model may include the following steps:

[0151] 301. When it is detected that there is a variable with a significance level greater than a preset significance level in the target model, the variable with a significance level greater than the preset significance level in the target model is determined as a variable to be deleted;

[0152] 302. Determine the variables in the target model whose significance level values ​​are not greater than the preset significance level values ​​as selected variables;

[0153] 303. Combining multiple groups of input variables according to the variables to be deleted and the selected variables, and inputting each group of input variables into the basic model to obtain multiple models to be screened, wherein the basic model is a model without variables.

[0154] Steps 301 to 303 are described in detail below.

[0155] In step 301, when it is detected that there are variables with a significance level greater than a preset significance level in the target model, the variables with a significance level greater than the preset significance level in the target model are determined as variables to be deleted.

[0156] Among them, if it is determined that the total number of variables in the target model is greater than the total number of preset variables, it can be detected whether there are variables in the target model whose significance level values ​​are greater than the preset significance level values. When it is detected that there are variables in the target model whose significance level values ​​are greater than the preset significance level values, the variables in the target model whose significance level values ​​are greater than the preset significance level values ​​are determined as variables to be deleted.

[0157] For example, the target model includes variables X1, X2, X3 and X4. If the significance level values ​​corresponding to variables X3 and X4 are greater than the preset significance level values, then X3 and X4 are determined as variables to be deleted.

[0158] In step 302, variables in the target model whose significance level values ​​are not greater than a preset significance level value are determined as selected variables.

[0159] For example, the target model includes variables X1, X2, X3 and X4. If the significance level values ​​corresponding to variables X1 and X2 are not greater than the preset significance level values, X1 and X2 are determined as variables to be deleted.

[0160] In step 303, multiple groups of input variables are combined according to the variables to be deleted and the selected variables, and each group of input variables is input into the basic model to obtain multiple models to be screened, where the basic model is a model without variables.

[0161] Finally, multiple groups of input variables are combined according to the variables to be deleted and the selected variables, and each group of input variables is input into the basic model to obtain multiple models to be screened, and the basic model is a model without variables. For example, the selected variables are determined to be selected and input into the basic model, then the variables to be deleted can be screened out and combined with the selected variables to obtain three groups of input variables, namely, one group of input variables X1, X2, X3 and X4, one group of input variables X1, X2 and X4, and one group of input variables X1, X2 and X3. Since the combination of input variables X1, X2, X3 and X4 is the variable combination of the current target model, this input variable combination is excluded, and one group of input variables X1, X2 and X4 and another group of input variables X1, X2 and X3 are selected and input into the basic model respectively, thereby obtaining two different models to be screened. Among them, the basic model is a model without variables.

[0162] It can be seen from steps 301 to 303 that in the present application, the selected variables in the target model whose variables are less than the preset significance level value are retained by presetting the significance level value, and then the variables to be deleted whose significance level values ​​in the target model are greater than the preset significance level value are combined with the selected variables. In this way, more input variables can be combined, thereby obtaining more models to be screened. Subsequently, the model with the highest evaluation value from the models to be screened is determined as the updated target model. In this way, the selected variables are used as fixed variables, and then the variables to be deleted are combined with them. In this way, the number of variable combinations can be reduced, thereby improving the efficiency of generating multiple groups of input variables.

[0163] In some embodiments, multiple models to be screened are combined according to each variable in the target model, including:

[0164] (1.1) When there is no variable in the target model with a significance level greater than the preset significance level, multiple groups of input variables are combined according to each variable in the target model;

[0165] (1.2) Input each set of input variables into the basic model to obtain multiple models to be screened.

[0166] For example, when the significance level value of each variable in the target model is less than or equal to the preset significance level value, but the total number of variables in the target model is greater than the preset total number of variables, some variables in the target model need to be deleted.

[0167] At this time, multiple groups of input variables can be combined according to each variable in the target model. For example, the variables in the target model include X1, X2, X3 and X4. The variables X1, X2, X3 and X4 can be arranged and combined to obtain multiple groups of input variables, and then the multiple groups of input variables are input into the basic model to obtain multiple models to be screened.

[0168] For example, variables X1 and X2 are combined to generate a set of input variables, and then the input variables are input into the basic model to generate a model to be screened. Variables X1, X2, and X3 are combined to generate a set of input variables, and then the input variables are input into the basic model to generate a model to be screened. Variables X3 and X4 are combined to generate a set of input variables, and then the input variables are input into the basic model to generate a model to be screened. By combining each variable in the target model, multiple sets of input variables are obtained, and then the variables are input into the basic model to obtain multiple models to be screened.

[0169] In this way, when there is no variable in the target model whose significance level value is greater than the preset significance level value, the combination of each variable in the target model can be exhausted, thereby obtaining multiple groups of input variables, and inputting them into the basic model to obtain multiple models to be screened. In this way, the model with the highest evaluation value can be accurately determined among the multiple models to be screened as the updated target model. The evaluation value of the target model determined in this way is high and more accurate, and the updated target model is determined as the current model. Subsequently, the model is updated based on the current model to obtain the updated target model. In this way, the updated target model can be updated on the basis of the current model with a higher evaluation value, thereby improving the update efficiency of the target model. Since the screening variables are obtained from the target model with the highest evaluation value among the multiple target models obtained historically, while improving the update efficiency of the target model, the efficiency of data screening is also indirectly improved.

[0170] In the above content, no matter how to obtain multiple models to be screened, the evaluation value of each model to be screened can be determined among the multiple models to be screened, and then the model with the highest evaluation value among the multiple models to be screened is determined as the updated target model according to the evaluation value. The updated target model is determined as the current model, and the execution is returned to select the selection variable from the multiple variables to be screened, and the current model is updated according to the selection variable to obtain the updated target model.

[0171] In this way, the process of dynamic change of the variables contained in the target model is realized, so as to try to continuously determine the selection variables and the variables already existing in the current model among the variables to be screened to update the current model and obtain the target model, thereby realizing the continuous acquisition of new target models.

[0172] In step 250, when it is detected that the target model meets the preset iteration condition, the target model with the highest evaluation value is selected from the historical target models, and the variables included in the target model with the highest evaluation value are determined as screening variables.

[0173] In the present application, the target model can be updated in the above manner to obtain multiple target models. However, the target model is not always updated without restriction. After each updated target model is obtained, it can be detected whether the target model meets the preset iteration conditions. If the target model meets the preset iteration conditions, the target model with the highest evaluation value is selected from the historical target models, and the variables contained in the target model with the highest evaluation value are determined as screening variables.

[0174] If the target model does not meet the preset iteration conditions, it is necessary to continue to obtain a new target model until the new target model meets the preset iteration conditions, and then select the target model with the highest evaluation value from the historical target models, and determine the variables contained in the target model with the highest evaluation value as the screening variables.

[0175] In some implementations, determining whether the target model meets a preset iteration condition may include:

[0176] (1.1) After each target model is obtained, the previous target model is obtained;

[0177] (1.2) When the variables included in the previous target model are the same as the variables included in the target model, it is determined that the target model meets the preset iteration conditions, and the target model with the highest evaluation value is selected from the historical target models, and the variables included in the target model with the highest evaluation value are determined as screening variables.

[0178] After each target model is obtained, the previous target model can be obtained, and then it is determined whether the variables contained in the previous target model are the same as the variables contained in the target model. When the variables contained in the previous target model are the same as the variables contained in the target model, it is determined that the target model meets the preset iteration conditions, and the target model with the highest evaluation value is selected from the historical target models, and the variables contained in the target model with the highest evaluation value are determined as screening variables.

[0179] For example, the current target model includes variables X1, X2, X3 and X4, and the previous target model includes variables X1, X2, X3 and X4. Then the variables included in the previous target model are the same as the variables included in the target model. At this time, it is determined that the target model meets the preset iteration conditions, and the target model with the highest evaluation value is selected from the historical target models, and the variables included in the target model with the highest evaluation value are determined as screening variables.

[0180] It should be noted that the process of target model change mainly includes the process of adding variables and the process of deleting variables. In the process of adding variables and deleting variables, the variables in the updated target model will change relative to the variables in the previous target model. If the variables contained in the previous target model are the same as the variables contained in the target model, it will occur during the transition between the process of adding variables and the process of deleting variables, that is, the currently updated target model will neither add nor reduce variables relative to the previous target model.

[0181] In some implementations, determining whether the target model meets a preset iteration condition may include:

[0182] (2.1) After each target model is obtained, the number of updates corresponding to the target model is determined;

[0183] (2.2) When the number of updates reaches a preset number of updates, it is determined that the target model meets the preset iteration conditions, and the target model with the highest evaluation value is selected from the historical target models, and the variables contained in the target model with the highest evaluation value are determined as screening variables.

[0184] Among them, after each target model is obtained, the corresponding update number of the target model can be determined. For example, if the target model is updated for the 500th time and the preset update number is 500 times, then it is determined that the update number of the target model reaches the preset update number, then it is determined that the target model meets the preset iteration conditions, and the target model with the highest evaluation value is selected from the target models obtained historically, and the variables contained in the target model with the highest evaluation value are determined as screening variables.

[0185] It can be understood that the target model with the highest evaluation value has the best model performance relative to other target models. In this case, the variables contained in the target model with the highest evaluation value can be determined as screening variables. In other words, these screening variables can improve the prediction results of the model, such as improving the model's distinguishing ability, and in financial service scenarios, it can distinguish whether a customer is suitable for handling corresponding financial services.

[0186] See also Figure 5 , Figure 5 It is a flowchart of determining screening variables provided in an embodiment of the present application.

[0187] In some implementations, selecting a target model with the highest evaluation value from historically obtained target models, and determining the variables included in the target model with the highest evaluation value as screening variables may include the following steps:

[0188] Step 401, determining a first target model whose total number of variables is not greater than a preset total number of variables from historically obtained target models;

[0189] Step 402: Determine a second target model in which the significance level value of each variable is not greater than a preset significance level value in the first target model;

[0190] Step 403: determine the target model with the highest evaluation value in the second target model, and determine the variable of the target model with the highest evaluation value as the screening variable.

[0191] Steps 401 to 403 are described in detail below.

[0192] In step 401, a first target model whose total number of variables is not greater than a preset total number of variables is determined from historically obtained target models.

[0193] In the present application, a preset total number of variables is specified, which is the maximum number of variables that can enter the basic model. Therefore, a first target model whose total number of variables is not greater than the preset total number of variables can be determined from the historical target models.

[0194] As can be seen from the above method of obtaining the target model, there are often multiple target models, so there are also multiple first target models whose total number of variables can be determined to be no greater than the total number of preset variables.

[0195] In step 402, a second target model is determined in the first target model in which the significance level value of each variable is not greater than a preset significance level value.

[0196] Then, a second target model is determined in the first target model in which the significance level value of each variable is not greater than the preset significance level value. Only when the significance level value of each variable is not greater than the preset significance level value, it is considered that the variable has a greater impact on the prediction results of the model.

[0197] For example, the variables included in one of the second target models are X1, X2, X3 and X4, and the significance level value of each variable in the variables X1, X2, X3 and X4 is less than the preset significance level value.

[0198] In step 403, a target model with the highest evaluation value is determined in the second target model, and the variable of the target model with the highest evaluation value is determined as a screening variable.

[0199] After determining the second target model, the evaluation value corresponding to each second target model, that is, the KS value of each second target model, can be determined, and then the variables included in the target model with the highest evaluation value can be determined, and these variables can be determined as screening variables.

[0200] Screening variables are variables determined from multiple models to be screened. Screening variables can be considered as variables that have a positive impact on the prediction results of the model and can improve the accuracy of the prediction results output by the model.

[0201] From step 401 to step 403, it can be seen that among the multiple target models obtained historically, the first target model that meets the total number of preset variables is first obtained, and then the second target model in which the significance level value of each variable is less than the preset significance level value is obtained from the first target model, and finally the target model with the highest evaluation value is determined from the second target model, so that the target model with the highest evaluation value finally selected satisfies the total number of variables less than the total number of preset variables, and the significance level value of each variable is less than the preset significance level value, then it is considered that the target model has higher performance, such as accurate prediction ability or discrimination ability. The prediction result of the target model with the highest evaluation value is caused by the variables input into the model, so these variables can be determined as screening variables, which greatly improves the efficiency of data screening in the variables to be screened.

[0202] In some embodiments, a third target model can be determined in the first target model, in which the number of variables whose significant level values ​​exceed the preset significant level value does not exceed two, that is, the third target model can allow a maximum of two variables whose significant level values ​​exceed the preset significant level value. Then, the target model with the highest evaluation value is determined in the third target model, and the variables of the target model with the highest evaluation value are determined as screening variables. In this way, the target model with the highest evaluation value can be determined in a larger range of the third target model, which is conducive to mining in the variables whose significant level values ​​exceed the preset significant level value, thereby determining the variables that are conducive to improving the accuracy of model prediction, which can improve the efficiency of data screening.

[0203] In this application, the screening variables are reasonably screened out in the above manner. When applied in actual scenarios later, these types of variables can be directly obtained and input into the model, so that the model outputs accurate prediction results. For example, in the field of financial services, if financial services need to be provided to an object, relevant data of the object can be obtained, which are all data of the same data type as the screening variables, and then the obtained data can be input into the model to output the corresponding score of the object. The score can reflect the credibility of the object in handling financial services, thereby avoiding financial risks caused by the instability of the object.

[0204] As can be seen from the above content, in an embodiment of the present application, by obtaining multiple variables to be screened, a selection variable is selected from the multiple variables to be screened, and the current model is updated according to the selection variable to obtain a target model; when it is detected that the target model does not meet the preset iteration condition, the first evaluation value corresponding to the current model and the second evaluation value corresponding to the target model are determined; when the total number of variables in the target model is not greater than the total number of preset variables, the second evaluation value is greater than the first evaluation value, and the significance level value of each variable in the target model is not greater than the preset significance level value, the target model is determined as the current model, and the process returns to execute to determine the selection variable from the multiple variables to be screened; when the total number is greater than the total number of preset variables, multiple models to be screened are combined according to each variable in the target model, and the model with the highest evaluation value among the multiple models to be screened is used as the updated target model, the updated target model is determined as the current model, and the process returns to execute to select the selection variable from the multiple variables to be screened; when it is detected that the target model meets the preset iteration condition, the target model with the highest evaluation value is selected from the target models obtained historically, and the variables contained in the target model with the highest evaluation value are determined as screening variables.

[0205] Thus, by setting the total number of preset variables of the variables of the model to be obtained, when the total number of variables of the target model is not greater than the total number of preset variables, the selected variables can be determined from multiple variables to be screened, and the target model is updated by the selected variables, so as to obtain a target model with a higher evaluation value than the previous model, so as to efficiently increase the variables in the target model; on the contrary, when the total number of variables in the target model is greater than the total number of preset variables, multiple models to be screened are combined according to each variable in the target model, and the model with the highest evaluation value among the multiple models to be screened is used as the updated target model, so as to efficiently reduce the redundant variables in the target model, obtain the updated target model, and then determine from the multiple variables to be screened. Determine the selection variables, and then update the target model by selecting the variables, so as to realize the continuous updating of the target model. In the process of updating the target model by adding or deleting variables, the target model with better performance than the current model can be continuously determined according to the evaluation value of the target model and the significance level value of each variable in the target model. Finally, when the target model meets the preset iteration conditions, the target model with the highest evaluation value can be determined from multiple target models obtained in history, and the variables contained in the target model with the highest evaluation value are used as the final screening variables. Compared with the related technology of realizing variable screening based on manual experience or traversal one by one, the efficiency of variable screening in this application is higher, thereby improving the data screening efficiency.

[0206] See also Figure 6 , Figure 6 It is another flowchart of the data screening method provided in an embodiment of the present application.

[0207] In order to briefly understand the technical solution in this application, please refer to Figure 6 , first obtain multiple variables to be screened and the current model. The current model has not been iterated yet, so the current model does not meet the preset iteration conditions. Then determine whether the total number of variables in the current model is greater than the total number of preset variables. If the total number of variables in the current model is not greater than the total number of preset variables, determine the selection variables from the variables to be screened, and input the selection variables into the current model to obtain the target model.

[0208] Then determine whether the target model meets the preset iteration conditions. If the target model does not meet the preset iteration conditions, determine whether the total number of variables in the target model is greater than the preset total number of variables. If the total number of variables in the target model is not greater than the preset total number of variables, determine the target model as the current model, determine the selection variables from the variables to be screened, and input the selection variables into the current model to obtain an updated target model.

[0209] If the updated target model does not meet the preset iteration conditions, and the total number of variables in the updated target model is not greater than the total number of preset variables, the target model is determined as the current model, the selection variables are determined from the variables to be screened, and the selection variables are input into the current model to obtain an updated target model.

[0210] If the updated target model does not meet the preset iteration conditions, and the total number of variables in the updated target model is greater than the preset total number of variables, it is necessary to partially delete the variables in the target model to obtain an updated target model.

[0211] If the updated target model does not meet the preset iteration conditions, and the total number of variables in the updated target model is not greater than the total number of preset variables, the target model is determined as the current model, the selection variables are determined from the variables to be screened, and the selection variables are input into the current model to obtain an updated target model.

[0212] By repeatedly executing the above method, the target model can be updated until the target model meets the preset iteration condition position, and then the target model with the highest evaluation value is determined from the historical target models, and the variables in the target model with the highest evaluation value are determined as the screening variables. Compared with the one-by-one traversal or manual screening in the related art, the data screening method in this application has higher data screening efficiency.

[0213] See also Figure 7 , Figure 7 is another flow chart of the data screening method provided in the embodiment of the present application. The data screening method may also include the following steps:

[0214] Step 501, obtaining multiple variables to be screened;

[0215] Step 502: Select a selection variable from a plurality of variables to be screened, and update the current model according to the selection variable to obtain a target model;

[0216] Step 503: when it is detected that the target model does not meet the preset iteration condition, determine the first evaluation value corresponding to the current model and the second evaluation value corresponding to the target model;

[0217] Step 504: when the total number of variables in the target model is not greater than the preset total number of variables, the second evaluation value is greater than the first evaluation value, and the significance level value of each variable in the target model is not greater than the preset significance level value, the target model is determined as the current model, and the process returns to determine the selected variable from the multiple variables to be screened;

[0218] Step 505: when the total number is greater than the preset total number of variables, and it is detected that there are variables with a significance level greater than the preset significance level in the target model, the variables with a significance level greater than the preset significance level in the target model are determined as variables to be deleted;

[0219] Step 506: Determine the variables in the target model whose significance level value is not greater than the preset significance level value as selected variables;

[0220] Step 507: combine multiple groups of input variables according to the variables to be deleted and the selected variables, and input each group of input variables into the basic model to obtain multiple models to be screened, where the basic model is a model without variables;

[0221] Step 508: taking the model with the highest evaluation value among the multiple models to be screened as the target model for updating;

[0222] Step 509: determine the updated target model as the current model, and return to execute selecting a selection variable from multiple variables to be screened;

[0223] Step 510: when it is detected that the target model satisfies the preset iteration condition, a first target model whose total number of variables is not greater than the preset total number of variables is determined from the historically obtained target models;

[0224] Step 511, determining a second target model in which the significance level value of each variable is not greater than a preset significance level value in the first target model;

[0225] Step 512: determine the target model with the highest evaluation value in the second target model, and determine the variable of the target model with the highest evaluation value as the screening variable.

[0226] In the above embodiments, the description of each embodiment has its own emphasis. For the part that is not described in detail in a certain embodiment, please refer to the detailed description of the above data screening method, which will not be repeated here.

[0227] See also Figure 8 , Figure 8 1 is a schematic diagram of the structure of a data screening device provided in an embodiment of the present application. The data screening device can execute the above-mentioned data screening method.

[0228] In the embodiments of the present application, the term "module" or "unit" refers to a computer program or a part of a computer program that has a predetermined function and works together with other related parts to achieve a predetermined goal, and can be implemented in whole or in part by using software, hardware (such as processing circuits or memories), or a combination thereof. Similarly, a processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be part of an overall module or unit that includes the function of the module or unit.

[0229] like Figure 8 As shown, the data screening device 600 includes:

[0230] The acquisition module 610 is used to acquire multiple variables to be screened, select a selection variable from the multiple variables to be screened, and update the current model according to the selection variable to obtain a target model;

[0231] A first determination module 620, configured to determine a first evaluation value corresponding to the current model and a second evaluation value corresponding to the target model when it is detected that the target model does not satisfy a preset iteration condition;

[0232] A second determination module 630 is used to determine the target model as the current model and return to execute determining a selected variable from a plurality of variables to be screened when the total number of variables in the target model is not greater than the total number of preset variables, the second evaluation value is greater than the first evaluation value, and the significance level value of each variable in the target model is not greater than the preset significance level value;

[0233] The third determination module 640 is used to combine multiple models to be screened according to each variable in the target model when the total number is greater than the total number of preset variables, and use the model with the highest evaluation value among the multiple models to be screened as the updated target model, determine the updated target model as the current model, and return to execute to select a selection variable from the multiple variables to be screened;

[0234] The screening module 650 is used to select the target model with the highest evaluation value from the historical target models when it is detected that the target model meets the preset iteration condition, and determine the variables included in the target model with the highest evaluation value as the screening variables.

[0235] In some implementations, the third determination module 640 includes a to-be-deleted submodule, a selection submodule, and a combination submodule;

[0236] A to-be-deleted submodule is used to, when detecting that there is a variable with a significance level value greater than a preset significance level value in the target model, determine the variable with a significance level value greater than the preset significance level value in the target model as a to-be-deleted variable;

[0237] A selection submodule is used to determine the variables whose significance level values ​​in the target model are not greater than the preset significance level values ​​as selected variables;

[0238] The combination submodule is used to combine multiple groups of input variables according to the variables to be deleted and the selected variables, and input each group of input variables into the basic model to obtain multiple models to be screened, and the basic model is a model without variables.

[0239] In some embodiments, the combination submodule is further used to:

[0240] When there is no variable in the target model whose significance level is greater than the preset significance level, multiple groups of input variables are combined according to each variable in the target model;

[0241] Each set of input variables is input into the basic model to obtain multiple models to be screened.

[0242] In some embodiments, the screening module 650 is used to:

[0243] Determine a first target model from the historically obtained target models, the total number of variables of which is not greater than the total number of preset variables;

[0244] Determine a second target model in which the significance level value of each variable is not greater than a preset significance level value in the first target model;

[0245] The target model with the highest evaluation value is determined in the second target model, and the variables of the target model with the highest evaluation value are determined as screening variables.

[0246] In some embodiments, the screening module 650 is further configured to:

[0247] After each target model is obtained, the previous target model is obtained;

[0248] When the variables included in the previous target model are the same as the variables included in the target model, it is determined that the target model meets the preset iteration conditions, and the target model with the highest evaluation value is selected from the historical target models, and the variables included in the target model with the highest evaluation value are determined as screening variables.

[0249] In some implementations, the second determining module 630 is further configured to:

[0250] When it is detected that the total number of variables in the target model is not greater than the preset total number of variables, if the second evaluation value is not greater than the first evaluation value, or the significance level value of any variable in the target model is greater than the preset significance level value, then return to execute to select a selected variable from multiple variables to be screened;

[0251] The current model is updated according to the selected variables to obtain the target model.

[0252] In some embodiments, the second determination module 630 further includes a detection submodule and a determination submodule;

[0253] A detection submodule, used for returning to execute selecting a selection variable from a plurality of variables to be screened when it is detected that there is a model identical to the target model in the historical target models before the target model is determined as the current model;

[0254] The determination submodule is used to determine the target model as the current model when there is no model identical to the target model in the target models obtained from the determination history.

[0255] In some embodiments, the acquisition module 610 includes an arrangement submodule and an acquisition submodule;

[0256] The arrangement submodule is used to determine the significance level value corresponding to each variable to be screened among the multiple variables to be screened before selecting the selected variable from the multiple variables to be screened;

[0257] Arrange each variable to be screened from low to high according to the significance level value to obtain a sequence of variables to be screened;

[0258] The multiple variables to be screened are processed in batches according to the sequence of the variables to be screened, so as to obtain multiple batches of variable sets to be screened.

[0259] Get the submodule to determine the set of variables to be screened corresponding to the current batch;

[0260] When there are unselected variables to be screened in the set of variables to be screened corresponding to the current batch, a selection variable is selected from the unselected variables to be screened;

[0261] When there are no unselected variables to be screened in the set of variables to be screened corresponding to the current batch, a selected variable is selected from the set of variables to be screened corresponding to the next batch.

[0262] In the above embodiments, the description of each embodiment has its own emphasis. For the part that is not described in detail in a certain embodiment, please refer to the detailed description of the above data screening method, which will not be repeated here.

[0263] As can be seen from the above content, in the embodiment of the present application, the acquisition module 610 acquires multiple variables to be screened, selects a selection variable from the multiple variables to be screened, and updates the current model according to the selection variable to obtain a target model; the first determination module 620 determines the first evaluation value corresponding to the current model and the second evaluation value corresponding to the target model when it is detected that the target model does not meet the preset iteration condition; the second determination module 630 determines the target model when the total number of variables in the target model is not greater than the preset total number of variables, the second evaluation value is greater than the first evaluation value, and the significance level value of each variable in the target model is not greater than the preset significance level value. The current model is returned to execute to determine the selection variable from multiple variables to be screened; when the total number is greater than the total number of preset variables, the third determination module 640 combines multiple models to be screened according to each variable in the target model, and uses the model with the highest evaluation value among the multiple models to be screened as the updated target model, determines the updated target model as the current model, and returns to execute to select the selection variable from multiple variables to be screened; when the screening module 650 detects that the target model meets the preset iteration conditions, it selects the target model with the highest evaluation value from the target models obtained historically, and determines the variables contained in the target model with the highest evaluation value as the screening variables.

[0264] Thus, by setting the total number of preset variables of the variables of the model to be obtained, when the total number of variables of the target model is not greater than the total number of preset variables, the selected variables can be determined from multiple variables to be screened, and the target model is updated by the selected variables, so as to obtain a target model with a higher evaluation value than the previous model, so as to efficiently increase the variables in the target model; on the contrary, when the total number of variables in the target model is greater than the total number of preset variables, multiple models to be screened are combined according to each variable in the target model, and the model with the highest evaluation value among the multiple models to be screened is used as the updated target model, so as to efficiently reduce the redundant variables in the target model, obtain the updated target model, and then determine from the multiple variables to be screened. Determine the selection variables, and then update the target model by selecting the variables, so as to realize the continuous updating of the target model. In the process of updating the target model by adding or deleting variables, the target model with better performance than the current model can be continuously determined according to the evaluation value of the target model and the significance level value of each variable in the target model. Finally, when the target model meets the preset iteration conditions, the target model with the highest evaluation value can be determined from multiple target models obtained in history, and the variables contained in the target model with the highest evaluation value are used as the final screening variables. Compared with the related technology of realizing variable screening based on manual experience or traversal one by one, the efficiency of variable screening in this application is higher, thereby improving the data screening efficiency.

[0265] The present application also provides a computer device, such as Fig. 9As shown, it shows a schematic diagram of the structure of a computer device provided in an embodiment of the present application, specifically:

[0266] The computer device may include a radio frequency (RF) circuit 701, a memory 702 including one or more computer-readable storage media, an input unit 703, a display unit 704, a sensor 705, an audio circuit 706, a wireless fidelity (WiFi) module 707, a processor 708 including one or more processing cores, and a power supply 709. Those skilled in the art will appreciate that Fig. 9 The terminal structure shown in the figure does not constitute a limitation on the terminal, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0267] The RF circuit 701 can be used for receiving and sending signals during the process of sending and receiving information or making calls. In particular, after receiving the downlink information of the base station, it is handed over to one or more processors 708 for processing; in addition, the data related to the uplink is sent to the base station. Generally, the RF circuit 701 includes but is not limited to an antenna, at least one amplifier, a tuner, one or more oscillators, a Subscriber Identity Module (SIM) card, a transceiver, a coupler, a low noise amplifier (LNA), a duplexer, etc. In addition, the RF circuit 701 can also communicate with the network and other devices through wireless communication. Wireless communication can use any communication standard or protocol, including but not limited to Global System of Mobile Communication (GSM), General Packet Radio Service (GPRS), Code Division Multiple Access (CDMA), Wideband Code Division Multiple Access (WCDMA), Long Term Evolution (LTE), email, Short Messaging Service (SMS), etc.

[0268] The memory 702 can be used to store software programs and modules. The processor 708 executes various functional applications and information retrieval by running the software programs and modules stored in the memory 702. The memory 702 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created according to the use of the terminal (such as audio data, a phone book, etc.), etc. In addition, the memory 702 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage devices. Accordingly, the memory 702 may also include a memory controller to provide the processor 708 and the input unit 703 with access to the memory 702.

[0269] The input unit 703 can be used to receive input digital or character information, and generate keyboard, mouse, joystick, optical or trackball signal input related to object setting and function control. Specifically, in a specific embodiment, the input unit 703 may include a touch-sensitive surface and other input devices. The touch-sensitive surface, also known as a touch display screen or a touch pad, can collect touch operations of an object on or near it (such as operations of an object using a finger, a stylus or any other suitable object or accessory on or near the touch-sensitive surface), and drive the corresponding connection device according to a pre-set program. Optionally, the touch-sensitive surface may include a touch detection device and a touch controller. Among them, the touch detection device detects the touch orientation of the object, detects the signal brought by the touch operation, and transmits the signal to the touch controller; the touch controller receives the touch information from the touch detection device, converts it into the touch point coordinates, and then sends it to the processor 708, and can receive and execute commands sent by the processor 708. In addition, the touch-sensitive surface can be implemented using multiple types such as resistive, capacitive, infrared and surface acoustic wave. In addition to the touch-sensitive surface, the input unit 703 may also include other input devices. Specifically, other input devices may include, but are not limited to, one or more of a physical keyboard, function keys (such as a volume control key, a switch key, etc.), a trackball, a mouse, a joystick, and the like.

[0270] The display unit 704 can be used to display information input by the object or information provided to the object and various graphic object interfaces of the terminal, which can be composed of graphics, text, icons, videos and any combination thereof. The display unit 704 may include a display panel. Optionally, the display panel can be configured in the form of a liquid crystal display (LCD), an organic light-emitting diode (OLED), etc. Further, the touch-sensitive surface can cover the display panel. When the touch-sensitive surface detects a touch operation on or near it, it is transmitted to the processor 708 to determine the type of touch event, and then the processor 708 provides a corresponding visual output on the display panel according to the type of touch event.

[0271] The terminal may also include at least one sensor 705, such as a light sensor, a motion sensor, and other sensors. Specifically, the light sensor may include an ambient light sensor and a proximity sensor, wherein the ambient light sensor may adjust the brightness of the display panel according to the brightness of the ambient light, and the proximity sensor may turn off the display panel and / or backlight when the terminal is moved to the ear. As a type of motion sensor, the gravity acceleration sensor can detect the magnitude of acceleration in each direction (generally three axes), and can detect the magnitude and direction of gravity when stationary. It can be used for applications that identify the posture of the mobile phone (such as horizontal and vertical screen switching, related games, magnetometer posture calibration), vibration recognition related functions (such as pedometer, tapping), etc.; As for other sensors that can be configured in the terminal, such as gyroscopes, barometers, hygrometers, thermometers, infrared sensors, etc., they will not be repeated here.

[0272] The audio circuit 706, the speaker, and the microphone can provide an audio interface between the object and the terminal. The audio circuit 706 can transmit the electrical signal converted from the received audio data to the speaker, which is converted into a sound signal for output; on the other hand, the microphone converts the collected sound signal into an electrical signal, which is received by the audio circuit 706 and converted into audio data, and then the audio data is output to the processor 708 for processing, and then sent to another terminal through the RF circuit 701, or the audio data is output to the memory 702 for further processing. The audio circuit 706 may also include an earplug jack to provide communication between an external headset and the terminal.

[0273] WiFi is a short-range wireless transmission technology. The terminal can help the object to send and receive emails, browse web pages and access streaming media through the WiFi module 707. It provides wireless broadband Internet access for the object. Fig. 9 A WiFi module 707 is shown, but it is understandable that it is not an essential component of the terminal and can be omitted as required without changing the essence of the invention.

[0274] The processor 708 is the control center of the terminal. It uses various interfaces and lines to connect various parts of the entire mobile phone. By running or executing software programs and / or modules stored in the memory 702 and calling data stored in the memory 702, it executes various functions of the terminal and processes data, thereby monitoring the mobile phone as a whole. Optionally, the processor 708 may include one or more processing cores; preferably, the processor 708 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, object interface and application programs, etc., and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 708.

[0275] The terminal also includes a power supply 709 (such as a battery) for supplying power to each component. Preferably, the power supply can be logically connected to the processor 708 through a power management system, so that the power management system can manage charging, discharging, and power consumption management. The power supply 709 can also include one or more DC or AC power supplies, recharging systems, power failure detection circuits, power converters or inverters, power status indicators, and other arbitrary components.

[0276] Although not shown, the terminal may also include a camera, a Bluetooth module, etc., which will not be described in detail here. Specifically in this embodiment, the processor 708 in the terminal will load the executable files corresponding to the processes of one or more applications into the memory 702 according to the following instructions, and the processor 708 will run the applications stored in the memory 702 to implement various functions:

[0277] Acquire multiple variables to be screened, select a selection variable from the multiple variables to be screened, and update the current model according to the selection variable to obtain a target model;

[0278] When it is detected that the target model does not meet the preset iteration condition, determining a first evaluation value corresponding to the current model and a second evaluation value corresponding to the target model;

[0279] When the total number of variables in the target model is not greater than the preset total number of variables, the second evaluation value is greater than the first evaluation value, and the significance level value of each variable in the target model is not greater than the preset significance level value, the target model is determined as the current model, and the process returns to determine the selected variable from the multiple variables to be screened;

[0280] When the total number is greater than the total number of preset variables, multiple models to be screened are combined according to each variable in the target model, and the model with the highest evaluation value among the multiple models to be screened is used as the updated target model, and the updated target model is determined as the current model, and the execution is returned to select the selected variable from the multiple variables to be screened;

[0281] When it is detected that the target model meets the preset iteration condition, the target model with the highest evaluation value is selected from the historical target models, and the variables included in the target model with the highest evaluation value are determined as screening variables.

[0282] In the above embodiments, the description of each embodiment has its own focus. For the part that is not described in detail in a certain embodiment, please refer to the detailed description of the data screening method above, which will not be repeated here.

[0283] As can be seen from the above content, in an embodiment of the present application, by obtaining multiple variables to be screened, a selection variable is selected from the multiple variables to be screened, and the current model is updated according to the selection variable to obtain a target model; when it is detected that the target model does not meet the preset iteration condition, the first evaluation value corresponding to the current model and the second evaluation value corresponding to the target model are determined; when the total number of variables in the target model is not greater than the total number of preset variables, the second evaluation value is greater than the first evaluation value, and the significance level value of each variable in the target model is not greater than the preset significance level value, the target model is determined as the current model, and the process returns to execute to determine the selection variable from the multiple variables to be screened; when the total number is greater than the total number of preset variables, multiple models to be screened are combined according to each variable in the target model, and the model with the highest evaluation value among the multiple models to be screened is used as the updated target model, the updated target model is determined as the current model, and the process returns to execute to select the selection variable from the multiple variables to be screened; when it is detected that the target model meets the preset iteration condition, the target model with the highest evaluation value is selected from the target models obtained historically, and the variables contained in the target model with the highest evaluation value are determined as screening variables.

[0284] Thus, by setting the total number of preset variables of the variables of the model to be obtained, when the total number of variables of the target model is not greater than the total number of preset variables, the selected variables can be determined from multiple variables to be screened, and the target model is updated by the selected variables, so as to obtain a target model with a higher evaluation value than the previous model, so as to efficiently increase the variables in the target model; on the contrary, when the total number of variables in the target model is greater than the total number of preset variables, multiple models to be screened are combined according to each variable in the target model, and the model with the highest evaluation value among the multiple models to be screened is used as the updated target model, so as to efficiently reduce the redundant variables in the target model, obtain the updated target model, and then determine from the multiple variables to be screened. Determine the selection variables, and then update the target model by selecting the variables, so as to realize the continuous updating of the target model. In the process of updating the target model by adding or deleting variables, the target model with better performance than the current model can be continuously determined according to the evaluation value of the target model and the significance level value of each variable in the target model. Finally, when the target model meets the preset iteration conditions, the target model with the highest evaluation value can be determined from multiple target models obtained in history, and the variables contained in the target model with the highest evaluation value are used as the final screening variables. Compared with the related technology of realizing variable screening based on manual experience or traversal one by one, the efficiency of variable screening in this application is higher, thereby improving the data screening efficiency.

[0285] A person of ordinary skill in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be completed by instructions, or by controlling related hardware through instructions. The instructions may be stored in a computer-readable storage medium and loaded and executed by a processor.

[0286] To this end, an embodiment of the present application provides a computer-readable storage medium, in which a plurality of instructions are stored, and the instructions can be loaded by a processor to execute the steps in any one of the data screening methods provided in the embodiments of the present application. For example, the instructions can execute the following steps:

[0287] Acquire multiple variables to be screened, select a selection variable from the multiple variables to be screened, and update the current model according to the selection variable to obtain a target model;

[0288] When it is detected that the target model does not meet the preset iteration condition, determining a first evaluation value corresponding to the current model and a second evaluation value corresponding to the target model;

[0289] When the total number of variables in the target model is not greater than the preset total number of variables, the second evaluation value is greater than the first evaluation value, and the significance level value of each variable in the target model is not greater than the preset significance level value, the target model is determined as the current model, and the process returns to determine the selected variable from the multiple variables to be screened;

[0290] When the total number is greater than the total number of preset variables, multiple models to be screened are combined according to each variable in the target model, and the model with the highest evaluation value among the multiple models to be screened is used as the updated target model, and the updated target model is determined as the current model, and the execution is returned to select the selected variable from the multiple variables to be screened;

[0291] When it is detected that the target model meets the preset iteration condition, the target model with the highest evaluation value is selected from the historical target models, and the variables included in the target model with the highest evaluation value are determined as screening variables.

[0292] In the above embodiments, the description of each embodiment has its own focus. For the part that is not described in detail in a certain embodiment, please refer to the detailed description of the data screening method above, which will not be repeated here.

[0293] According to one aspect of the present application, a computer program product or a computer program is provided, the computer program product or the computer program includes a computer instruction, and the computer instruction is stored in a computer-readable storage medium. A processor of a computer device reads the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the data screening method provided in various optional implementations provided in the above embodiments.

[0294] The specific implementation of the above operations can be found in the previous embodiments, which will not be described in detail here.

[0295] The computer-readable storage medium may include: a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0296] Since the instructions stored in the computer-readable storage medium can execute the steps in any data screening method provided in the embodiments of the present application, the beneficial effects that can be achieved by any data screening method provided in the embodiments of the present application can be achieved. Please refer to the previous embodiments for details and will not be repeated here.

[0297] The above is a detailed introduction to a data screening method, device, equipment and medium provided in the embodiments of the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for technical personnel in this field, according to the ideas of the present application, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A method for processing financial service data, characterized in that: The financial service data processing method is applied to a computer device, and the method comprises: Acquire multiple variables to be screened of the financial service object, wherein the multiple variables to be screened correspond to multiple features of the financial service object respectively, and different variables to be screened correspond to different data types, select a selection variable from the multiple variables to be screened, and update the current financial scorecard model according to the selection variable to obtain a target financial scorecard model; When it is detected that the target financial scorecard model does not meet the preset iteration condition, determining a first Kolmogorov-Smirnov value corresponding to the current financial scorecard model and a second Kolmogorov-Smirnov value corresponding to the target financial scorecard model; When the total number of variables in the target financial scorecard model is not greater than the preset total number of variables, the second Kolmogorov-Smirnov value is greater than the first Kolmogorov-Smirnov value, and the significance level value of each variable in the target financial scorecard model is not greater than the preset significance level value, the target financial scorecard model is determined as the current financial scorecard model, and the process of determining a selected variable from the multiple variables to be screened is returned; When the total number is greater than the total number of preset variables, a plurality of financial scorecard models to be screened are combined according to each variable in the target financial scorecard model, and the model with the highest Kolmogorov-Smirnov value among the plurality of financial scorecard models to be screened is used as the updated target financial scorecard model, the updated target financial scorecard model is determined as the current financial scorecard model, and the execution is returned to select a selected variable from the plurality of variables to be screened; When it is detected that the target financial scorecard model satisfies the preset iteration condition, the target financial scorecard model with the highest Kolmogorov-Smirnov value is selected from the historically obtained target financial scorecard models, and the variables included in the target financial scorecard model with the highest Kolmogorov-Smirnov value are determined as screening variables; Determining a target feature of the same data type as the screening variable from the features of the financial service object, the target feature being composed of at least one feature corresponding to the financial service object, and the target feature being used to improve the scoring accuracy of the target financial scoring card model with the highest Kolmogorov-Smirnov value; The target feature is input into the target financial scoring card model with the highest Kolmogorov-Smirnov value to obtain the financial scoring value of the object, and whether to provide corresponding financial services to the object is determined according to the financial scoring value. The financial scoring value is used to evaluate the credit credibility of the financial service object.

2. The financial service data processing method according to claim 1, characterized in that: The step of combining multiple financial scorecard models to be screened according to each variable in the target financial scorecard model comprises: When it is detected that there is a variable with a significance level greater than the preset significance level in the target financial scorecard model, the variable with a significance level greater than the preset significance level in the target financial scorecard model is determined as a variable to be deleted; Determine the variable whose significance level value is not greater than the preset significance level value in the target financial scorecard model as a selected variable; A plurality of groups of input variables are combined according to the variables to be deleted and the selected variables, and each group of input variables is input into a basic financial scorecard model to obtain a plurality of financial scorecard models to be screened, wherein the basic financial scorecard model is a financial scorecard model without variables.

3. The financial service data processing method according to claim 2, characterized in that: The step of combining multiple financial scorecard models to be screened according to each variable in the target financial scorecard model comprises: When there is no variable in the target financial scorecard model whose significance level value is greater than the preset significance level value, combining multiple groups of input variables according to each variable in the target financial scorecard model; Each group of input variables is input into the basic financial scorecard model to obtain multiple financial scorecard models to be screened.

4. The financial service data processing method according to claim 1, characterized in that: The target financial scoring card model with the highest Kolmogorov-Smirnov value is selected from the target financial scoring card models obtained historically, and the variables included in the target financial scoring card model with the highest Kolmogorov-Smirnov value are determined as screening variables, including: Determine a first target financial scorecard model whose total number of variables is not greater than the total number of preset variables from the target financial scorecard models obtained historically; Determining a second target financial scorecard model in which the significance level value of each variable is not greater than the preset significance level value in the first target financial scorecard model; A target financial scoring card model with the highest Kolmogorov-Smirnov value is determined in the second target financial scoring card model, and the variables of the target financial scoring card model with the highest Kolmogorov-Smirnov value are determined as screening variables.

5. The financial service data processing method according to claim 1, characterized in that: Also includes: After obtaining the target financial scorecard model each time, obtaining the previous target financial scorecard model; When the variables included in the previous target financial scoring card model are the same as the variables included in the target financial scoring card model, it is determined that the target financial scoring card model meets the preset iteration condition, and the target financial scoring card model with the highest Kolmogorov-Smirnov value is selected from the historical target financial scoring card models, and the variables included in the target financial scoring card model with the highest Kolmogorov-Smirnov value are determined as screening variables.

6. The financial service data processing method according to claim 1, characterized in that: Also includes: After obtaining the target financial scorecard model each time, determining the number of updates corresponding to the target financial scorecard model; When the number of updates reaches a preset number of updates, it is determined that the target financial scoring card model meets the preset iteration condition, and the target financial scoring card model with the highest Kolmogorov-Smirnov value is selected from the target financial scoring card models obtained historically, and the variables included in the target financial scoring card model with the highest Kolmogorov-Smirnov value are determined as screening variables.

7. The financial service data processing method according to claim 1, characterized in that: Also includes: When it is detected that the total number of variables in the target financial scorecard model is not greater than the preset total number of variables, if the second Kolmogorov-Smirnov value is not greater than the first Kolmogorov-Smirnov value, or the significance level value of any variable in the target financial scorecard model is greater than the preset significance level value, then returning to execute selecting a selected variable from the multiple variables to be screened; The current financial scorecard model is updated according to the selected variables to obtain a target financial scorecard model.

8. The financial service data processing method according to claim 1, characterized in that: Before determining the target financial scorecard model as the current financial scorecard model, the method further includes: When it is detected that there is a model identical to the target financial scorecard model in the historical target financial scorecard model, returning to execute selecting a selection variable from the multiple variables to be screened; The step of determining the target financial scorecard model as the current financial scorecard model includes: When it is determined that there is no model identical to the target financial scorecard model among the historically obtained target financial scorecard models, the target financial scorecard model is determined as the current financial scorecard model.

9. The financial service data processing method according to claim 1, characterized in that: Before selecting a selection variable from the plurality of variables to be screened, the method further includes: Among the multiple variables to be screened, determining a significance level value corresponding to each variable to be screened; Arrange each of the variables to be screened from low to high according to the significance level value to obtain a sequence of variables to be screened; The multiple variables to be screened are processed in batches according to the sequence of variables to be screened to obtain multiple batches of variable sets to be screened.

10. The financial service data processing method according to claim 9, characterized in that: The step of selecting a selection variable from the plurality of variables to be screened includes: Determine the set of variables to be screened corresponding to the current batch; When there are unselected variables to be screened in the set of variables to be screened corresponding to the current batch, a selected variable is selected from the unselected variables to be screened; When there are no unselected variables to be screened in the set of variables to be screened corresponding to the current batch, a selected variable is selected from the set of variables to be screened corresponding to the next batch.

11. A financial service data processing device, characterized in that: The financial service data processing device is applied to a computer device, and the device comprises: an acquisition module, configured to acquire a plurality of variables to be screened of a financial service object, wherein the plurality of variables to be screened respectively correspond to a plurality of features of the financial service object, and different variables to be screened correspond to different data types, select a selection variable from the plurality of variables to be screened, and update the current financial scorecard model according to the selection variable to obtain a target financial scorecard model; A first determination module, configured to determine a first Kolmogorov-Smirnov value corresponding to the current financial scorecard model and a second Kolmogorov-Smirnov value corresponding to the target financial scorecard model when it is detected that the target financial scorecard model does not meet a preset iteration condition; a second determination module, configured to determine the target financial scorecard model as the current financial scorecard model, and return to execute the step of determining a selected variable from the multiple variables to be screened, when the total number of variables in the target financial scorecard model is not greater than the total number of preset variables, the second Kolmogorov-Smirnov value is greater than the first Kolmogorov-Smirnov value, and the significance level value of each variable in the target financial scorecard model is not greater than the preset significance level value; A third determination module is used for, when the total number is greater than the total number of preset variables, combining multiple financial scorecard models to be screened according to each variable in the target financial scorecard model, and taking the model with the highest Kolmogorov-Smirnov value among the multiple financial scorecard models to be screened as the updated target financial scorecard model, determining the updated target financial scorecard model as the current financial scorecard model, and returning to execute to select a selected variable from the multiple variables to be screened; A screening module, configured to select a target financial scoring card model with the highest Kolmogorov-Smirnov value from the historically obtained target financial scoring card models when it is detected that the target financial scoring card model satisfies a preset iteration condition, and determine the variables included in the target financial scoring card model with the highest Kolmogorov-Smirnov value as screening variables; A scoring module, configured to determine a target feature of the same data type as the screening variable from the features of the financial service object, wherein the target feature is composed of at least one feature corresponding to the financial service object, and the target feature is used to improve the scoring accuracy of the target financial scoring card model with the highest Kolmogorov-Smirnov value; The target feature is input into the target financial scoring card model with the highest Kolmogorov-Smirnov value to obtain the financial scoring value of the object, and whether to provide corresponding financial services to the object is determined according to the financial scoring value. The financial scoring value is used to evaluate the credit credibility of the financial service object.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor to execute the financial service data processing method according to any one of claims 1 to 10.

13. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the financial service data processing method according to any one of claims 1 to 10 is implemented.

14. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the financial service data processing method according to any one of claims 1 to 10 is implemented.

Images