Limit estimation method and device, computer equipment and storage medium

Abstract

The invention belongs to the field of financial information, and relates to a limit estimation method and device, computer equipment and a storage medium. The quota estimation comprises the following steps: acquiring historical transaction information of a customer, and extracting historical flow information of the customer from the historical transaction information of the customer; feature construction is carried out on multiple dimensions according to the customer historical flow information, and flow derivative features including flow change trend information are obtained; performing feature screening on the basis of the running water derivative features to obtain running water in-mold features; inputting the flow-into-model feature into a pre-trained credit line prediction model to obtain a credit line model score; calculating a credit basic limit according to the customer historical flow information and the flow change trend information; and obtaining a credit line range, and performing credit line estimation according to the credit extension basic line and the credit line model to obtain an effective credit line. According to the invention, the effective credit line of the customer can be accurately estimated according to the historical transaction information of the customer, and the estimation efficiency is improved.

Description

Technical Field

[0001] The present application relates to the field of financial information, and in particular to a credit limit estimation method, device, computer equipment and storage medium. Background Art

[0002] In the vast field of bank credit business, accurately assessing the credit capacity of borrowers is the core link to ensure the safety of funds and promote financial stability. At present, the technical means generally relied on by the industry mainly focus on using the data collected by the corporate credit system of the People's Bank of China, combined with publicly available financial information such as financial statements and tax records, to preliminarily judge the repayment ability and risk level of the enterprise, and then set a reasonable credit limit. However, this traditional method has gradually become limited in the face of an increasingly complex and changing financial market.

[0003] What is particularly noteworthy is that for “white household” enterprises or individuals that have never had a credit record, due to the lack of historical data support, traditional assessment models often find it difficult to effectively capture their true credit status, causing banks to be caught in a dilemma in credit decisions, worrying about the risks caused by excessive credit and missing out on high-quality customers. In addition, even for enterprises with a credit history, the complexity and diversity of their financial data, coupled with the ever-changing market environment, have led to significant fluctuations in the operating conditions and capital liquidity of enterprises in different economic cycles. These factors have greatly increased the difficulty of determining credit limits, and traditional static analysis methods are difficult to fully and accurately reflect the actual debt repayment capacity of enterprises, which in turn affects the efficiency and risk control capabilities of bank credit business. Summary of the invention

[0004] The purpose of the embodiments of the present application is to provide a credit limit estimation method, apparatus, computer equipment and storage medium to solve the problem of being unable to quickly and accurately estimate a customer's credit limit.

[0005] In order to solve the above technical problems, the present application embodiment provides a credit limit estimation method, which adopts the following technical solution:

[0006] Obtaining historical transaction information of a customer, and extracting historical transaction information of the customer from the historical transaction information of the customer;

[0007] Constructing features based on the historical transaction information of the customer to obtain transaction-derived features;

[0008] Perform feature screening based on the pipeline derived features to obtain pipeline input features;

[0009] Inputting the customer's historical transaction information and the flow model features into a pre-trained credit limit prediction model to obtain a credit limit model score;

[0010] Calculate the basic credit limit based on the customer's historical transaction information and the transaction trend information in the transaction derivative feature;

[0011] The credit limit is estimated based on the basic credit limit and the credit limit model to obtain an effective credit limit.

[0012] Furthermore, the step of obtaining the customer's historical transaction information and extracting the customer's historical transaction information from the customer's historical transaction information specifically includes:

[0013] Get the transaction information extraction identifier;

[0014] Extracting the customer's historical transaction information from a database according to the transaction information extraction identifier;

[0015] Performing data cleaning on the customer's historical transaction information to obtain standard customer's historical transaction information;

[0016] The standard customer historical transaction information is parsed to obtain the customer historical flow information.

[0017] Furthermore, the step of constructing features based on the customer's historical transaction information to obtain transaction-derived features specifically includes:

[0018] Analyze the customer's historical transaction information to obtain transaction amount information, transaction frequency information, transaction category information, transaction counterparty information, and change trend information;

[0019] Based on feature aggregation and feature conversion, feature construction is performed on the transaction amount information, the transaction frequency information, the transaction category information, the transaction counterparty information, and the change trend information to obtain the flow-derived features.

[0020] Furthermore, the transaction flow derived features include transaction amount features, transaction frequency features, transaction type features, transaction counterparty features, and transaction flow change trend information. The step of performing feature screening based on the transaction flow derived features to obtain transaction flow modeling features specifically includes:

[0021] Performing feature correlation analysis based on the flow-derived features to obtain flow feature correlation information;

[0022] Performing feature information analysis based on the flow-derived features to obtain flow feature information;

[0023] Performing feature stability analysis based on the flow-derived features to obtain flow feature stability information;

[0024] The derived flow features are grouped and screened according to the flow feature relevance, the flow feature information volume, and the flow feature stability to obtain the flow input model features.

[0025] Furthermore, before the step of inputting the customer's historical transaction information into the pre-trained credit limit prediction model to obtain the credit limit model score, the following steps are also included:

[0026] Obtain sample customer credit information;

[0027] Preprocessing the sample customer credit information to obtain standardized sample customer credit information;

[0028] Extracting features from the sample customer credit information to obtain sample customer transaction features;

[0029] The standardized sample customer credit information and the sample customer transaction characteristics are input into a preset gradient boosting decision tree model for training to obtain the credit limit prediction model.

[0030] Furthermore, the transaction flow derived characteristics include transaction amount characteristics, transaction frequency characteristics, transaction type characteristics, transaction counterparty characteristics, and transaction flow change trend information. The step of calculating the basic credit limit based on the customer's historical transaction flow information and the transaction flow change trend information specifically includes:

[0031] Extracting the flow change trend information from the flow derived features, and performing standard quantification on the customer's historical flow information and the flow change trend information to obtain the customer's historical flow volume and the customer's flow change volume;

[0032] Obtain a preset basic credit limit calculation formula, and calculate the basic credit limit according to the basic credit limit calculation formula. The basic credit limit calculation formula is:

[0033] α=γ(FLx)+(1-γ)(FLy)

[0034] Among them, α is the basic credit limit, FLx is the change in customer turnover, FLy is the customer's historical turnover, and γ is the ratio of the customer's change in turnover to the customer's historical turnover, that is, FLx / FLy.

[0035] Furthermore, the step of estimating the credit limit based on the basic credit limit and the credit limit model to obtain an effective credit limit specifically includes:

[0036] Obtaining information of cooperative financial institutions, and obtaining a credit limit range based on the information of the cooperative financial institutions;

[0037] A preset credit limit calculation formula is obtained, and a credit limit calculation is performed according to the credit limit calculation formula to obtain a calculated credit limit, and the calculated credit limit is verified by the credit limit range, and the verified calculated credit limit is used as the effective credit limit. The credit limit calculation formula is:

[0038] Q=α / (LK*M)

[0039] Among them, Q is the effective credit limit, α is the basic credit limit, L is the first constant, K is the second constant, and M is the credit limit model score.

[0040] In order to solve the above technical problems, the embodiment of the present application further provides a credit limit estimation device, which adopts the following technical solution:

[0041] An information acquisition module is used to acquire the customer's historical transaction information and extract the customer's historical transaction information from the customer's historical transaction information;

[0042] An information construction module, used to construct features based on the customer's historical transaction information to obtain transaction-derived features;

[0043] An information screening module, used for screening features based on the pipeline derived features to obtain pipeline input features;

[0044] A model prediction module, used for inputting the customer's historical transaction information and the flow input model features into a pre-trained credit limit prediction model to obtain a credit limit model score;

[0045] An information generation module, used for calculating the basic credit limit according to the customer's historical transaction information and the transaction change trend information in the transaction derivative feature;

[0046] The credit limit estimation module is used to estimate the credit limit based on the basic credit limit and the credit limit model to obtain an effective credit limit.

[0047] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0048] A computer device includes a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the credit estimation method as described in any one of the above items are implemented.

[0049] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0050] A computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the steps of the credit estimation method as described in any one of the above items are implemented.

[0051] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0052] This embodiment obtains the customer's historical transaction information, and extracts the customer's historical flow information from the customer's historical transaction information; constructs features based on the customer's historical flow information to obtain flow derivative features; performs feature screening based on the flow derivative features to obtain flow input features; inputs the customer's historical transaction information and the flow input features into the pre-trained credit line prediction model to obtain the credit line model score; calculates the basic credit line based on the customer's historical flow information and the flow change trend information in the flow derivative features; estimates the credit line based on the basic credit line and the credit line model score to obtain the effective credit line. This embodiment can effectively estimate the customer's effective credit line based on the customer's historical transaction information and improve the efficiency of the estimation. This makes it easier for banks to provide customers with appropriate and effective lending services. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0054] Figure 1 A flow chart of an embodiment of a method for estimating a credit limit according to the present application;

[0055] Figure 2 yes Figure 1 A flowchart of a specific implementation of step S10;

[0056] Figure 3 yes Figure 1 A flowchart of a specific implementation of step S20;

[0057] Figure 4 yes Figure 1 A flowchart of a specific implementation of step S30;

[0058] Figure 5 yes Figure 1 A flowchart of a specific implementation of step S50;

[0059] Figure 6 yes Figure 1A flowchart of a specific implementation of step S60;

[0060] Figure 7 is a structural schematic diagram of an embodiment of a credit limit estimation device according to the present application;

[0061] Figure 8 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0062] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0063] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0064] 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 accompanying drawings.

[0065] Continue to refer Figure 1 , shows a flow chart of an embodiment of a credit estimation method according to the present application. The credit estimation method comprises the following steps:

[0066] Step S10, obtaining the customer's historical transaction information, and extracting the customer's historical transaction information from the customer's historical transaction information;

[0067] In this embodiment, customer historical transaction information includes deposit information, income and expenditure flow information, credit history information, counterparty information, related party information, etc. Customer historical flow information refers to the inflow (such as income, deposits, refunds, etc.) and outflow (such as expenditure, transfer, withdrawal, loan repayment, etc.) of funds for all transaction activities in the past period of time. The customer in the customer historical transaction information and customer historical flow information can refer to an individual user or an enterprise. In this embodiment, the customer historical transaction information refers to the enterprise's historical deposit information, income and expenditure flow information, credit history information, counterparty information, and related party information. Among them, deposit information includes the account information opened by the enterprise in various banks, account balance change records, time deposits and current deposits, capital flow direction and scale, etc.; income and expenditure flow information includes sales revenue, service fees, raw material procurement, salary payment, tax payment, etc.; credit history information includes the enterprise's past loan records, repayment status, guarantee measures, credit rating, etc.; counterparty information includes transaction records between the enterprise and its partners, suppliers, customers, etc.; related party transaction information includes information when the enterprise conducts transactions with its parent company, subsidiaries, other enterprises controlled by the same party, or other enterprises with other related relationships.

[0068] Step S20, constructing features based on the customer's historical transaction information to obtain transaction-derived features;

[0069] In this embodiment, feature construction includes feature aggregation and feature conversion. By aggregating features from different dimensions such as time, amount, transaction nature, and counterparty, commonalities are extracted from multiple historical transaction information of customers, and aggregated into transaction-derived features of customers in this dimension, including the change trend of transaction in the time dimension; through feature conversion, the structure of customer transaction information is changed, and the information is converted into structured transaction-derived features that are easier to be interpreted by the model.

[0070] Step S30, performing feature screening based on the pipeline derived features to obtain pipeline input features;

[0071] In this embodiment, the feature screening step includes feature correlation analysis, feature information analysis, and feature stability analysis. Through feature correlation analysis, the intrinsic correlation between features is found and divided into different feature groups; through feature information analysis, the effectiveness of features for prediction is judged, and features with stronger prediction effectiveness are selected from the features with higher correlation in each group; through feature stability analysis, features with low stability and poor generalization ability are discarded to improve the stability of the model.

[0072] Step S40, inputting the customer's historical transaction information and the flow input model features into a pre-trained credit limit prediction model to obtain a credit limit model score;

[0073] In this embodiment, the pre-trained credit limit prediction model adopts the LightGBM algorithm model. By setting the parameters of the LightGBM model, such as the number of trees (n_estimators), the maximum depth of each tree (max_depth), the learning rate (learning_rate), etc., the model is trained based on the obtained sample customer credit information and the sample customer transaction features extracted from the sample customer credit information, thereby effectively obtaining the pre-trained credit limit prediction model.

[0074] Step S50, calculating the basic credit limit according to the customer's historical transaction information and the transaction change trend information in the transaction derivative feature;

[0075] In this embodiment, the basic credit limit is estimated according to a preset basic limit calculation formula, and the preset basic limit calculation formula can be extracted from the database according to the first calculation formula extraction identifier. In this embodiment, the customer's historical transaction flow information refers to the customer's transaction flow data in the past year, and the flow change trend information includes the flow trend direction: indicating whether the flow volume is increasing or decreasing, that is, an upward trend or a downward trend. Flow change amplitude: indicates the size or degree of change in flow volume. Flow periodicity: If the flow changes show obvious periodicity, such as peaks or valleys at fixed time nodes every month or quarter. Flow stability: Evaluate the stability of flow changes, that is, whether there are sudden drastic fluctuations or long-term stable growth / decrease trends. Seasonal factors: If the customer's business is greatly affected by seasonal factors, then the flow changes will show obvious seasonal characteristics.

[0076] Step S60, estimating the credit limit according to the basic credit limit and the credit limit model to obtain an effective credit limit.

[0077] In this embodiment, the credit limit range refers to the credit limit range of the bank's credit business, which is determined by the bank based on its own risk tolerance, capital status, market strategy, customer credit status and other factors. The credit limit range sets the maximum and minimum limits that the bank is willing to provide to customers of specific types or conditions. The effective credit limit refers to the most appropriate credit limit determined after considering various conditions of the customer within the credit limit range.

[0078] This embodiment obtains the customer's historical transaction information, and extracts the customer's historical flow information from the customer's historical transaction information; constructs features based on the customer's historical flow information to obtain flow derivative features including flow change trend information; performs feature screening based on the flow derivative features to obtain flow input features; inputs the customer's historical transaction information and the flow input features into the pre-trained credit line prediction model to obtain the credit line model score; calculates the basic credit line based on the customer's historical flow information and the flow change trend information in the flow derivative features; estimates the credit line based on the basic credit line and the credit line model score to obtain the effective credit line. This embodiment can effectively estimate the customer's effective credit line based on the customer's historical transaction information and improve the efficiency of the estimation. This facilitates banks to provide customers with appropriate and effective lending services.

[0079] Continue to refer Figure 2 In some optional implementations of this embodiment, step S10 includes the following steps:

[0080] Step S101, obtaining a transaction information extraction identifier;

[0081] In this embodiment, the transaction information extraction identifier is a unique identifier corresponding to the customer's historical transaction information. The transaction information extraction identifier can be a data ID or UUID (universally unique identifier). In this embodiment, since the enterprise is the customer entity, the transaction information extraction identifier can use the unified social credit code corresponding to the enterprise.

[0082] Step S102, extracting the customer's historical transaction information from a database according to the transaction information extraction identifier;

[0083] In this embodiment, a matching query is performed in the database through the transaction information extraction identifier, and the matched information is extracted, thereby obtaining the customer's historical transaction information.

[0084] Step S103, performing data cleaning on the customer historical transaction information to obtain standard customer historical transaction information;

[0085] In this embodiment, data cleaning includes processing missing values: if there are few missing values ​​and they have little impact on the analysis results, records with missing values ​​can be directly deleted; use statistical values ​​(such as mean, median, mode) or business rules to fill missing values ​​(for time series data, interpolation methods can be used to estimate missing values). Processing outliers: identify outliers through statistical methods (such as standard deviation, IQR interquartile range method) or business rules; according to the nature of the outliers, you can choose to delete, correct or retain the outliers, and consider their impact on the analysis results. Remove duplicate records: compare the values ​​of all fields, and records that are exactly the same are considered duplicate records; only compare the values ​​of some key fields, such as product number, transaction date, transaction amount, etc.; delete duplicate records or merge information in duplicate records according to business needs. Format standardization Unify the format of date and time; unify the precision and unit of numerical values, such as retaining two decimal places for currency amounts; remove spaces, special characters, and redundant punctuation marks in text fields. Data type conversion: convert string type values ​​to numerical types for mathematical operations; convert date and time strings to date and time types for time series analysis. Data verification: Ensure that the cleaned data is logically consistent, such as the relationship between transaction amount and transaction quantity; confirm that key fields are not missing and the integrity of product transaction data is preserved.

[0086] Step S104, parsing the standard customer historical transaction information to obtain the customer historical flow information.

[0087] In this embodiment, the steps of parsing the standard customer historical transaction information include: transaction classification: classifying the transaction records according to the transaction type (such as income, expenditure, transfer, etc.) to facilitate the subsequent analysis of the flow of different categories; information alignment: sorting out the relationship between the information from different data tables to form the data table lineage of customer-account-flow, so as to facilitate the subsequent cross-table analysis of customer flow. By performing the above parsing steps on the standard customer historical transaction information, the customer historical flow information is obtained.

[0088] This embodiment obtains a transaction information extraction identifier; extracts the customer's historical transaction information from a database according to the transaction information extraction identifier; performs data cleansing on the customer's historical transaction information to obtain standard customer's historical transaction information; and parses the standard customer's historical transaction information, thereby effectively obtaining standard and effective customer historical transaction information.

[0089] Continue to refer Figure 3 In some optional implementations of this embodiment, step S20 includes the following steps:

[0090] Step S201, parsing the customer's historical transaction information to obtain transaction amount information, transaction frequency information, transaction category information, transaction counterparty information, and change trend information;

[0091] In this embodiment, the transaction amount information and transaction frequency information are parsed from the customer's historical transaction information through statistical analysis methods, and then the transaction type and transaction counterparty of the customer's historical transaction information are identified to obtain the transaction category information and transaction counterparty information. Time series analysis is performed on time series data such as transaction amount and transaction number to identify long-term trends and cyclical fluctuations and obtain change trend information.

[0092] Step S202, based on feature aggregation and feature conversion, feature construction is performed on the transaction amount information, the transaction frequency information, the transaction category information, the transaction counterparty information, and the change trend information to obtain the flow-derived features.

[0093] In this embodiment, according to the principal component analysis method, feature aggregation is adopted to aggregate different dimensions such as time, amount, nature of transaction, counterparty, etc. to extract commonalities from multiple historical transaction information of customers, and aggregate them into the customer's transaction-derived features on this dimension. The feature aggregation methods include grouping statistics such as total, median, average, extreme value, standard deviation, frequency, etc. according to multiple dimensions such as time, region, industry, etc.; then, feature conversion is used to change the structure of the customer's transaction information, and convert the information into structured transaction-derived features that are easier to be interpreted by the model. Among them, the feature conversion methods include four arithmetic operations, linear combination, ranking coding, string truncation, time interval, etc.

[0094] This embodiment performs feature aggregation and feature conversion on the customer's historical transaction information to obtain multi-dimensional and interpretable transaction-derived features.

[0095] Continue to refer Figure 4 In some optional implementations of this embodiment, step S30 includes the following steps:

[0096] Step S301, performing feature correlation analysis based on the flow-derived features to obtain flow-feature correlation information;

[0097] In this embodiment, the step of feature correlation analysis includes: calculating the correlation between each feature through flow-derived features, which can be achieved through a correlation coefficient matrix; according to the correlation matrix, identifying which features are highly correlated, thereby obtaining flow feature correlation information, wherein highly correlated features may reflect the same underlying factors or information redundancy.

[0098] Step S302, performing feature information analysis based on the flow-derived features to obtain flow feature information;

[0099] In this embodiment, the step of feature information analysis includes: analyzing the information content of the feature through the discriminant effect of a single feature in the flow-derived feature on the customer's historical flow information on the target variable (in this embodiment, the credit default rate), and the feature with high information content has stronger predictive power for the model; for features that are too strong, it is necessary to check their distribution to see whether overfitting is caused by the distribution of customer historical flow information; for the monotonicity of feature effectiveness, the grouping effectiveness should be made monotonic as much as possible or there should be a single peak / valley value, so that the feature has better interpretability.

[0100] Step S303, performing feature stability analysis based on the flow-derived feature to obtain flow feature stability information;

[0101] In this embodiment, the steps of feature stability analysis include: calculating the population stability index PSI (Population Stability Index) of the flow-derived features over time; grouping the flow-derived features according to the discrimination between positive and negative samples, and observing the relative stability of the discrimination between samples in different time periods after grouping, which is intuitively manifested as whether there is an intersection between the line graphs of each group.

[0102] Step S304, grouping and screening the derived flow features according to the flow feature relevance, the flow feature information volume, and the flow feature stability to obtain the flow input model features.

[0103] In this embodiment, the feature with the strongest explanatory power for the target variable (usually the overdue rate in this patent scenario) is selected from the highly correlated customer flow features, and then some unstable flow features are eliminated according to the PSI value of the feature to obtain the flow input model feature.

[0104] This embodiment performs feature construction processing on the customer's historical transaction information through feature aggregation and feature conversion to obtain customer transaction derived features; performs feature correlation, information volume and stability analysis based on the customer transaction derived features to obtain transaction feature correlation, information volume and stability information; performs feature screening based on the transaction feature correlation, information volume and stability information, thereby obtaining transaction input features that enable the model to effectively interpret transaction changes, so as to facilitate subsequent model construction processing.

[0105] In some optional implementations of this embodiment, before step S40, the following steps are further included:

[0106] Obtain sample customer credit information;

[0107] In this embodiment, the sample customer credit information refers to credit records extracted from historical customer credit information, and can be obtained from a database of a bank, financial institution or other relevant data source.

[0108] Preprocessing the sample customer credit information to obtain standardized sample customer credit information;

[0109] In this embodiment, preprocessing includes data cleaning: removing duplicate records, processing missing values ​​(such as by interpolation, deletion or filling default values), and correcting outliers (such as transaction amounts that are obviously beyond common sense); data standardization / normalization: since the value ranges and units of different features may be different, in order to improve the efficiency and accuracy of model training, it is usually necessary to standardize or normalize the features so that all features are at the same order of magnitude; data conversion: converting categorical data (such as transaction type) into numerical data (such as using one-hot encoding or label encoding).

[0110] Extracting features from the sample customer credit information to obtain sample customer transaction features;

[0111] In this embodiment, the sample customer transaction characteristics are sample model characteristics obtained by feature screening of the sample customer credit information through the above step S304, and the sample model characteristics include information on transaction frequency, transaction amount, transaction type, transaction object, balance change, transaction time, transaction periodicity, risk record, etc.

[0112] The standardized sample customer credit information and the sample customer transaction characteristics are input into a preset gradient boosting decision tree model for training to obtain the credit limit prediction model.

[0113] In this embodiment, the preset gradient boosting decision tree model adopts the LightGBM algorithm model. The LightGBM algorithm model can effectively predict the credit limit based on the provided sample customer transaction characteristics to improve the accuracy of the predicted data.

[0114] This embodiment obtains sample customer credit information; pre-processes the sample customer transaction information to obtain standardized sample customer credit information; extracts features from the sample customer credit information to obtain sample customer transaction features; inputs the standardized sample customer credit information and the sample customer transaction features into a preset gradient boosting decision tree model for training, thereby effectively obtaining a credit limit prediction model that can accurately predict the credit limit, so as to facilitate subsequent credit limit prediction.

[0115] Continue to refer Figure 5 In some optional implementations of this embodiment, step S50 includes the following steps:

[0116] Step S501, extracting the flow change trend information from the flow derived features, and performing standard quantification on the customer's historical flow information and the flow change trend information to obtain the customer's historical flow volume and the customer's flow change volume;

[0117] In this embodiment, the standard quantification of the customer's historical transaction information can be achieved by normalizing the data. Normalization refers to scaling the data of the customer's historical transaction information to between 0 and 1. The normalized transaction amount = (transaction amount - minimum transaction amount) / (maximum transaction amount - minimum transaction amount). The standard quantification of the flow change trend information includes calculating the absolute change value or relative change rate of the flow volume in adjacent time periods. Absolute change value: Δ flow volume = flow volume in the current time period - flow volume in the previous time period; relative change rate: change rate = flow volume in the previous time period Δ flow volume × 100%. Time period division: Calculate the difference in the total transaction amount or number of transactions between the current time period and the previous time period based on the absolute change rate or relative change rate to obtain the turnover change for the time period; obtain the corresponding transaction type weight value and transaction frequency weight value according to different transaction types and transaction frequencies, and perform weighted calculation on the turnover change through the transaction type weight value and transaction frequency weight value to obtain the customer turnover change, where the transaction type weight value and transaction frequency weight value are extracted through the weight value mapping table, which records information on evaluating the customer's value or behavior pattern based on different transaction types (such as purchase, return, consultation, etc.) and transaction frequencies (such as high frequency, medium frequency, low frequency).

[0118] Step S502, obtaining a preset basic credit limit calculation formula, and calculating the basic credit limit according to the basic credit limit calculation formula, the basic credit limit calculation formula is:

[0119] α=γ(FLx)+(1-γ)(FLy)

[0120] Among them, α is the basic credit limit, FLx is the change in customer turnover, FLy is the customer's historical turnover, and γ is the ratio of the customer's change in turnover to the customer's historical turnover, that is, FLx / FLy.

[0121] In this embodiment, various parameters are brought into the formula for calculation, thereby effectively obtaining the calculation result of the basic credit limit.

[0122] This embodiment obtains the customer's historical flow amount and the customer's flow change amount by standardizing the customer's historical flow information and the flow change trend information; obtains a preset basic credit calculation formula, and calculates the basic credit amount according to the basic credit calculation formula, and the basic credit calculation formula is: α=γ(FLx)+(1-γ)(FLy) where α is the basic credit amount, FLx is the customer's flow change amount, FLy is the customer's historical flow amount, and γ is the ratio of the customer's flow change amount to the customer's historical flow amount, that is, FLx / FLy. This effectively realizes the calculation of the basic credit amount to facilitate the subsequent credit line estimation.

[0123] Continue to refer Figure 6 In some optional implementations of this embodiment, step S60 includes the following steps:

[0124] Step S601, obtaining cooperative financial institution information, and obtaining a credit limit range according to the cooperative financial institution information;

[0125] In this embodiment, the cooperative financial institution information refers to the financial institution information of the credit business partner, including the needs of the partner financial institution. The corresponding cooperative financial institution object is determined through the cooperative financial institution information, and the needs of the partner financial institution are obtained from the cooperative financial institution object to obtain the credit limit range, where the credit limit range can be a specific numerical range, indicating the upper and lower limits of the credit limit that a bank or financial institution is willing to provide for this type of business.

[0126] Step S602, obtaining a preset credit limit calculation formula, and performing credit limit calculation according to the credit limit calculation formula to obtain a calculated credit limit, and verifying the calculated credit limit through the credit limit range, and taking the verified calculated credit limit as the effective credit limit, the credit limit calculation formula is:

[0127] Q=α / (LK*M)

[0128] Among them, Q is the effective credit limit, α is the basic credit limit, L is the first constant, K is the second constant, and M is the credit limit model score.

[0129] In this embodiment, L is the first constant, K is the second constant, L and K are determined by the credit limit range expected by the bank, for example, the credit limit range is 200,000-2 million, the number of sample credit customers is 10,000, then by adjusting L and K, the minimum Q value of 10,000 sample credit customers is 200,000, the maximum Q value is 2 million, and the remaining Q values ​​fall between 200,000-2 million. (Some extreme value customers can be ignored according to actual conditions, such as the lowest 5% and the highest 5%, so that 90% of customers are within this range), and the values ​​of L and K are set by the above method. In this embodiment, M is the model output score of the thousand-point system, the value of L is preset to 30, and the value of K is preset to 0.01, which can be adjusted according to actual conditions. After calculating and determining the effective credit limit formula, by comparing the effective credit limit of the sample customer with the credit limit range, it is verified whether the effective credit limit falls within the credit limit range, so as to judge whether the effective credit limit formula complies with the bank's credit rules, so as to ensure the real-time validity of the effective credit limit formula.

[0130] This embodiment obtains customer account information according to the customer historical transaction information; obtains the credit limit range according to the needs of the cooperative financial institution; obtains a preset credit limit calculation formula, and calculates the credit limit according to the credit limit calculation formula to obtain a calculated credit limit, and verifies the calculated credit limit formula through the credit limit range, and uses the verified calculated credit limit formula as the effective credit limit formula, and the credit limit calculation formula is: Q = α / (LK * M) where Q is the effective credit limit, α is the credit basic limit, L is the first constant, K is the second constant, and M is the credit limit model score. Thereby, the effective credit limit is accurately estimated to improve the accuracy of customer credit limit prediction.

[0131] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0132] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0133] Further references Figure 7 , as a response to the above Figure 1 In order to realize the method shown in the figure, the present application provides an embodiment of a credit estimation device, and the embodiment of the device is Figure 1 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0134] like Figure 7 As shown, the credit limit estimation device 700 described in this embodiment includes: an information acquisition module 701, an information construction module 702, an information screening module 703, a model prediction module 704, an information generation module 705, and a credit limit estimation module 706. Among them:

[0135] The information acquisition module 701 is used to acquire the customer's historical transaction information and extract the customer's historical transaction information from the customer's historical transaction information;

[0136] An information construction module 702 is used to construct features based on the customer's historical transaction information to obtain transaction-derived features;

[0137] An information screening module 703 is used to perform feature screening based on the pipeline derived features to obtain pipeline input features;

[0138] Model prediction module 704, used to input the customer's historical transaction information and the flow input model features into the pre-trained credit limit prediction model to obtain the credit limit model score;

[0139] The information generation module 705 is used to calculate the basic credit limit according to the customer's historical transaction information and the transaction change trend information in the transaction derivative feature;

[0140] The credit limit estimation module 706 is used to obtain a credit limit range, perform credit limit estimation based on the basic credit limit and the credit limit model, and obtain an effective credit limit.

[0141] By adopting the above system, this embodiment can obtain the customer's historical transaction information, and extract the customer's historical flow information from the customer's historical transaction information; perform feature construction based on the customer's historical flow information to obtain flow derivative features including flow change trend information; perform feature screening based on the flow derivative features to obtain flow input features; input the customer's historical transaction information and the flow input features into the pre-trained credit line prediction model to obtain the credit line model score; calculate the basic credit line based on the customer's historical flow information and the flow change trend information in the flow derivative features; perform credit line estimation based on the basic credit line and the credit line model score to obtain the effective credit line. This embodiment can effectively estimate the customer's effective credit line based on the customer's historical transaction information and improve the efficiency of estimation. This makes it convenient for banks to provide customers with appropriate and effective lending services.

[0142] To solve the above technical problems, the present application also provides a computer device. Figure 8 , Figure 8 This is a basic structural block diagram of the computer device in this embodiment.

[0143] The computer device 8 includes a memory 81, a processor 82, and a network interface 83 that are interconnected through a system bus. It should be noted that the figure only shows a computer device 8 with components 81-83, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0144] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0145] The memory 81 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 81 can be an internal storage unit of the computer device 8, such as a hard disk or memory of the computer device 8. In other embodiments, the memory 81 can also be an external storage device of the computer device 8, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (FlashCard), etc. equipped on the computer device 8. Of course, the memory 81 can also include both the internal storage unit of the computer device 8 and its external storage device. In this embodiment, the memory 81 is generally used to store the operating system and various application software installed on the computer device 8, such as computer-readable instructions of the credit estimation method, etc. In addition, the memory 81 can also be used to temporarily store various types of data that have been output or are to be output.

[0146] The processor 82 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments. The processor 82 is generally used to control the overall operation of the computer device 8. In this embodiment, the processor 82 is used to run computer-readable instructions or process data stored in the memory 81, such as computer-readable instructions for running the credit estimation method.

[0147] The network interface 83 may include a wireless network interface or a wired network interface. The network interface 83 is generally used to establish a communication connection between the computer device 8 and other electronic devices.

[0148] By adopting the above-mentioned computer equipment, this embodiment can obtain the customer's historical transaction information, and extract the customer's historical flow information from the customer's historical transaction information; perform feature construction based on the customer's historical flow information to obtain flow derivative features including flow change trend information; perform feature screening based on the flow derivative features to obtain flow input features; input the customer's historical transaction information and the flow input features into the pre-trained credit line prediction model to obtain the credit line model score; calculate the basic credit line based on the customer's historical flow information and the flow change trend information in the flow derivative features; perform credit line estimation based on the basic credit line and the credit line model score to obtain the effective credit line. This embodiment can effectively estimate the customer's effective credit line based on the customer's historical transaction information and improve the efficiency of estimation. This makes it convenient for banks to provide customers with appropriate and effective lending services.

[0149] The present application also provides another implementation, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the credit estimation method as described above.

[0150] This embodiment, by using the above-mentioned computer-readable storage medium, can obtain the customer's historical transaction information, and extract the customer's historical flow information from the customer's historical transaction information; perform feature construction based on the customer's historical flow information to obtain flow derivative features including flow change trend information; perform feature screening based on the flow derivative features to obtain flow input features; input the customer's historical transaction information and the flow input features into the pre-trained credit line prediction model to obtain the credit line model score; calculate the basic credit line based on the customer's historical flow information and the flow change trend information in the flow derivative features; perform credit line estimation based on the basic credit line and the credit line model score to obtain the effective credit line. This embodiment can effectively estimate the customer's effective credit line based on the customer's historical transaction information and improve the efficiency of estimation. This makes it convenient for banks to provide customers with appropriate and effective lending services.

[0151] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0152] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A quota estimation method, characterized in that: include: Obtaining historical transaction information of a customer, and extracting historical transaction information of the customer from the historical transaction information of the customer; Perform feature construction based on the customer's historical transaction information to obtain transaction-derived features; Perform feature screening based on the pipeline derived features to obtain pipeline input features; Inputting the customer's historical transaction information and the flow-in model features into a pre-trained credit limit prediction model to obtain a credit limit model score; Calculate the basic credit limit based on the customer's historical transaction information and the transaction trend information in the transaction derivative feature; The credit limit is estimated based on the basic credit limit and the credit limit model to obtain an effective credit limit.

2. The credit estimation method according to claim 1, characterized in that: The step of obtaining the customer's historical transaction information and extracting the customer's historical transaction information from the customer's historical transaction information specifically includes: Get the transaction information extraction identifier; Extracting the customer's historical transaction information from a database according to the transaction information extraction identifier; Performing data cleaning on the customer's historical transaction information to obtain standard customer's historical transaction information; The standard customer historical transaction information is parsed to obtain the customer historical flow information.

3. The credit estimation method according to claim 1, characterized in that: The step of constructing features based on the customer's historical transaction information to obtain transaction-derived features specifically includes: Analyze the customer's historical transaction information to obtain transaction amount information, transaction frequency information, transaction category information, transaction counterparty information, and change trend information; Based on feature aggregation and feature conversion, feature construction is performed on the transaction amount information, the transaction frequency information, the transaction category information, the transaction counterparty information, and the change trend information to obtain the flow-derived features.

4. The credit estimation method according to claim 1, characterized in that: The derived transaction characteristics include transaction amount characteristics, transaction frequency characteristics, transaction type characteristics, transaction counterparty characteristics, and transaction trend information. The step of performing feature screening based on the derived transaction characteristics to obtain transaction modeling characteristics specifically includes: Performing feature correlation analysis based on the flow-derived features to obtain flow feature correlation information; Performing feature information analysis based on the flow-derived features to obtain flow feature information; Performing feature stability analysis based on the flow-derived features to obtain flow feature stability information; The derived flow features are grouped and screened according to the flow feature relevance, the flow feature information volume, and the flow feature stability to obtain the flow input model features.

5. The credit estimation method according to claim 1, characterized in that: Before the step of inputting the customer historical transaction information and the flow model features into the pre-trained credit limit prediction model to obtain the credit limit model score, the following steps are also included: Obtain sample customer credit information; Preprocessing the sample customer credit information to obtain standardized sample customer credit information; Extracting features from the sample customer credit information to obtain sample customer transaction features; The standardized sample customer credit information and the sample customer transaction characteristics are input into a preset gradient boosting decision tree model for training to obtain the credit limit prediction model.

6. The credit limit estimation method according to claim 1, characterized in that: The transaction flow derived characteristics include transaction amount characteristics, transaction frequency characteristics, transaction type characteristics, transaction counterparty characteristics, and transaction flow change trend information. The step of calculating the basic credit limit based on the customer's historical transaction flow information and the transaction flow change trend information specifically includes: Extracting the flow change trend information from the flow derived features, and performing standard quantification on the customer's historical flow information and the flow change trend information to obtain the customer's historical flow volume and the customer's flow change volume; Obtain a preset basic credit limit calculation formula, and calculate the basic credit limit according to the basic credit limit calculation formula. The basic credit limit calculation formula is: α=γ(FLx)+(1-γ)(FLy) Among them, α is the basic credit limit, FLx is the change in customer turnover, FLy is the customer's historical turnover, and γ is the ratio of the customer's change in turnover to the customer's historical turnover, that is, FLx / FLy.

7. The credit estimation method according to claim 1, characterized in that: The step of estimating the credit limit based on the basic credit limit and the credit limit model to obtain an effective credit limit specifically includes: Obtaining information of cooperative financial institutions, and obtaining a credit limit range based on the information of the cooperative financial institutions; A preset credit limit calculation formula is obtained, and a credit limit calculation is performed according to the credit limit calculation formula to obtain a calculated credit limit, and the calculated credit limit is verified by the credit limit range, and the verified calculated credit limit is used as the effective credit limit. The credit limit calculation formula is: Q=α / (LK*M) Among them, Q is the effective credit limit, α is the basic credit limit, L is the first constant, K is the second constant, and M is the credit limit model score.

8. A credit estimation device, characterized in that: include: An information acquisition module is used to acquire the customer's historical transaction information and extract the customer's historical transaction information from the customer's historical transaction information; An information construction module, used to construct features based on the customer's historical transaction information to obtain transaction-derived features; An information screening module, used for screening features based on the pipeline derived features to obtain pipeline input features; A model prediction module, used for inputting the customer's historical transaction information and the flow input model features into a pre-trained credit limit prediction model to obtain a credit limit model score; An information generation module, used for calculating the basic credit limit according to the customer's historical transaction information and the transaction change trend information in the transaction derivative feature; The credit limit estimation module is used to estimate the credit limit based on the basic credit limit and the credit limit model to obtain an effective credit limit.

9. A computer device, characterized in that: The method comprises a memory and a processor, wherein the memory stores computer-readable instructions, and the processor implements the steps of the credit estimation method according to any one of claims 1 to 7 when executing the computer-readable instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the steps of the credit estimation method according to any one of claims 1 to 7 are implemented.

Images