A risk warning and disposal method for post-loan risks in automobile finance

Abstract

The present invention relates to the field of automobile finance technology, and specifically to a post-loan risk early warning disposal method for automobile finance, comprising the following steps: based on the post-loan transaction information of the borrower, collecting and analyzing the borrower's bank statement and consumption record information, including the amount, time and frequency of multiple transactions, to generate borrower bill data. In the present invention, through the borrower's consumption record and bank statement, the transaction object, time and amount of multiple transactions are analyzed, risky transaction behaviors are identified, default risks are identified at an early stage, the value of mortgage assets is evaluated in combination with the market trend of the vehicle and the borrower's credit record, the accuracy of asset evaluation is improved, and the overdue repayment period is adjusted according to the borrower's actual repayment ability by calculating the repayment risk score and matching the collection notice, so as to improve the fund recovery rate and the financial stability of loan services, optimize the risk management efficiency, reduce the losses caused by default, and protect the interests of consumers and investors.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile finance, and in particular to a method for early warning and handling of automobile finance post-loan risks. Background Art

[0002] The field of automotive finance technology focuses on promoting the purchase, sale and leasing of automobiles through financial tools, services and technologies, including vehicle loans, lease agreements, insurance services and credit management. In automotive finance, by providing financing support for individual and corporate car purchase needs, including credit assessment, asset management, credit risk assessment and customer relationship management, the purchase and management of vehicles are promoted, and by managing the repurchase and loan default risks associated with automobile transactions, the repayment of loans and the safety of funds are ensured, the efficiency of capital use is optimized, and financial stability and market competitiveness are improved.

[0003] Among them, the post-loan risk warning and disposal method of auto finance aims to realize the post-monitoring and management risks of auto loans, and to achieve effective prevention and control of loan risks by identifying signals that may lead to defaults at an early stage. By analyzing various factors such as the borrower's repayment behavior, changes in economic conditions, and market environment, it warns of default risks and provides corresponding disposal strategies, including readjusting repayment plans, implementing collection measures, and taking legal actions, to help financial institutions reduce losses, maintain asset value, and protect the interests of consumers and investors.

[0004] Traditional auto finance post-loan risk warning and disposal methods lack the ability to track and analyze borrowers' transaction behaviors in real time. The lack of analysis of borrowers' bank statements and consumption records makes it difficult to detect abnormal transaction behaviors in a timely manner, and it is impossible to effectively intervene in the early stages of default risk formation. In terms of mortgage asset valuation, it is difficult to consider the immediate changes in market trends, resulting in inaccurate asset value assessments, affecting the value of the loan's collateral and the effectiveness of risk control. It performs poorly when the market environment changes rapidly, and is unable to effectively adapt to fluctuations in borrowers' economic conditions and changes in market conditions, resulting in financial institutions facing higher financial risks. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a post-loan risk early warning and disposal method for automobile finance.

[0006] In order to achieve the above-mentioned purpose, the present invention adopts the following technical scheme: a method for early warning and handling of automobile finance post-loan risk, comprising the following steps:

[0007] S1: Based on the borrower's post-loan transaction information, collect and analyze the borrower's bank statements and consumption record information, including the amount, time and frequency of multiple transactions, to generate the borrower's bill data;

[0008] S2: Based on the borrower's bill data, by analyzing the transaction objects, time, and amount of multiple transactions, abnormal transaction information and risky transaction behaviors are identified, and transaction behavior analysis results are generated;

[0009] S3: Based on the transaction behavior analysis results, the value of the vehicle is evaluated by analyzing the market trend, vehicle condition and historical usage data of the borrower's vehicle, and the mortgage asset evaluation results are generated in combination with the borrower's credit record and economic status;

[0010] S4: Using the mortgage asset evaluation results, by analyzing the asset status of multiple borrowers, combined with historical repayment behaviors and overdue records, calculating repayment risk scores, and generating risk borrower identification results;

[0011] S5: matching collection notification information for borrowers of multiple risk levels according to the risk borrower identification results, adjusting the frequency of sending notifications, recording the borrower's repayment response behavior, and generating a collection response record;

[0012] S6: Using the collection response record, taking into account the borrower's financial status and capital recovery efficiency, adjusting the loan repayment period and overdue interest rate and calculating the repayment amount, and generating a post-loan risk processing result.

[0013] As a further solution of the present invention, the borrower's billing data includes transaction date information, transaction object information, and transaction amount information; the transaction behavior analysis results include abnormal transaction identification records, risky transaction type data, and risky transaction period information; the mortgage asset assessment results include real-time automobile market value assessment data, expected depreciation rate, maintenance and repair records; the risky borrower identification results include credit risk rating, predicted default rate, and borrower financial stress assessment results; the collection response records include response time analysis results, repayment amount change information, and repayment willingness assessment information; the post-loan risk processing results include repayment plan adjustment results, overdue interest rate update results, and expected funds recovery efficiency.

[0014] As a further solution of the present invention, based on the borrower's post-loan transaction information, the borrower's bank statements and consumption record information are collected and analyzed, including the amount, time and frequency of multiple transactions, and the steps of generating the borrower's bill data are specifically as follows:

[0015] S101: Based on the borrower's post-loan transaction information, collect the borrower's bank statements and consumption records, record the transaction date, amount and frequency information, and generate transaction data collection results;

[0016] S102: Based on the transaction data collection result, clean and format the data, identify duplicate records and erroneous inputs, and generate consumption record processing results;

[0017] S103: Based on the consumption record processing result, the transaction data is classified, the total amount and transaction frequency of various transactions are calculated, and the borrower's bill data is generated.

[0018] As a further solution of the present invention, based on the borrower's bill data, by analyzing the transaction objects, time, and amount of multiple transactions, abnormal transaction information and risky transaction behaviors are identified, and the steps of generating transaction behavior analysis results are specifically as follows:

[0019] S201: Based on the borrower's bill data, by analyzing the amounts and transaction times of multiple transactions and comparing them with preset monitoring thresholds, abnormal transaction amounts and transaction times are identified, and abnormal transaction comparison results are generated;

[0020] S202: Based on the abnormal transaction comparison result, the background information of the abnormal transaction is analyzed, including the payment location and the transaction object, and the risky transactions are classified and recorded to generate a risky transaction classification result;

[0021] S203: Based on the risky transaction classification result, by evaluating the impact of various transaction behaviors on the borrower's credit status, a logistic regression algorithm is used to calculate the credit impact scores of the various transaction behaviors to generate a transaction behavior analysis result.

[0022] As a further solution of the present invention, the logistic regression algorithm is according to the formula:

[0023]

[0024] Calculate the borrower's default probability, where p is the probability of the borrower's default, b0 is the intercept term, b1 is the weight related to the transaction amount, b2 is the weight related to the transaction frequency, b3 is the weight related to the transaction time interval, b4 is the weight related to the borrower's historical overdue times, b5 is the weight related to the amount of funds flowing into the borrower's account, x1 is the transaction amount, x2 is the transaction frequency, x3 is the transaction time interval, x4 is the borrower's historical overdue times, x5 is the amount of funds flowing into the borrower's account, and e is the base of the natural logarithm.

[0025] As a further solution of the present invention, based on the transaction behavior analysis results, the value of the vehicle is evaluated by analyzing the market trend, vehicle condition and historical usage data of the borrower's vehicle, and the steps of generating the mortgage asset evaluation results in combination with the borrower's credit record and economic status are specifically as follows:

[0026] S301: Based on the transaction behavior analysis results, the borrower's vehicle information is collected, basic conditions and historical usage data of the vehicle are recorded, including vehicle model, vehicle age, mileage and maintenance records, and a vehicle condition data record is generated;

[0027] S302: Based on the vehicle condition data record, analyze the sales prices and demand trends of multiple vehicles through real-time market data to generate market trend analysis results;

[0028] S303: Based on the market trend analysis results, the borrower's vehicle value and change trend are evaluated in real time, and the mortgage asset evaluation results are generated in combination with the borrower's credit record and economic status.

[0029] As a further solution of the present invention, the steps of using the mortgage asset evaluation results to calculate the repayment risk score by analyzing the asset status of multiple borrowers and combining historical repayment behaviors and overdue records to generate risk borrower identification results are as follows:

[0030] S401: Using the mortgage asset evaluation results, collect and record asset status data and historical repayment behavior records of multiple borrowers, including overdue records and repayment frequency information, to generate asset and behavior data;

[0031] S402: Based on the asset and behavior data, analyze the borrower's historical repayment behavior and overdue record, evaluate the stability of the repayment behavior and the probability of overdue payment, and generate a stability analysis result;

[0032] S403: Based on the stability analysis results, the repayment risk scores of multiple borrowers are calculated, the asset value and credit status are considered, the overdue risk level is evaluated, and the risk borrower identification results are generated.

[0033] As a further solution of the present invention, according to the risk borrower identification result, the collection notification information is matched for borrowers of multiple risk levels, the notification sending frequency is adjusted, the borrower's repayment response behavior is recorded, and the steps of generating the collection response record are specifically as follows:

[0034] S501: According to the risk borrower identification result, matching the collection notice content according to the risk levels of multiple borrowers, including the notification information and the sending frequency, to generate a collection notice matching result;

[0035] S502: Based on the collection notice matching result, collection notice information is sent to multiple borrowers, and the repayment behavior of the borrowers after receiving the notice is recorded, including the time and amount of repayment, and collection information response data is generated;

[0036] S503: Analyze the collection information response data, use the K-means clustering algorithm, analyze the borrower's response speed and repayment completion rate, evaluate the effects of various collection notice contents, and generate a collection response record.

[0037] As a further solution of the present invention, the K-means clustering algorithm is based on the formula:

[0038]

[0039] Calculate the clustering effect of the collection strategy, where S is the total internal variance, k is the number of clusters, and C i is the set of data points in the i-th cluster, x time is the borrower’s response time data point, x amount is the borrower’s response amount data point, x behavior is the borrower’s repayment behavior data point, μ time is the center point of the time cluster, μ amount is the center point of the amount cluster, μ behavior is the center point of the behavior cluster, w1 is the time weight coefficient, w2 is the amount weight coefficient, w3 is the behavior weight coefficient, i is the cluster index, and x is a single data point.

[0040] As a further solution of the present invention, the collection response record is used to consider the borrower's financial status and capital recovery efficiency, adjust the loan repayment period and overdue interest rate and calculate the repayment amount, and the steps of generating the post-loan risk processing result are specifically as follows:

[0041] S601: Analyze the collection response record, analyze the repayment amount and loan interest rate of multiple loan contracts, and compare them with the actual repayment ability of multiple borrowers to identify the overdue risks of multiple borrowers and generate overdue risk identification data;

[0042] S602: Based on the overdue risk identification data, considering the fund recovery efficiency and the borrower's financial capacity, identifying the repayment period and overdue interest rate that need to be adjusted, and generating repayment condition adjustment data;

[0043] S603: Adjust the data according to the repayment conditions, adjust the borrower's repayment contract, monitor the adjusted repayment behavior and analyze the impact on repayment efficiency, and generate post-loan risk processing results.

[0044] Compared with the prior art, the advantages and positive effects of the present invention are:

[0045] In the present invention, the transaction objects, time and amount of multiple transactions are analyzed through the borrower's consumption records and bank statements, risky transaction behaviors are identified, default risks are identified at an early stage, the value of the mortgaged assets is evaluated in combination with the market trend of the vehicle and the borrower's credit record, the accuracy of asset evaluation is improved, and the overdue repayment period is adjusted according to the borrower's actual repayment ability by calculating the repayment risk score and matching the collection notice, thereby improving the fund recovery rate and the financial stability of the loan service, optimizing the risk management efficiency, reducing the losses caused by default, and protecting the interests of consumers and investors. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a schematic diagram of the main steps of the present invention;

[0047] Figure 2 It is a schematic diagram of the refinement of S1 of the present invention;

[0048] Figure 3 It is a schematic diagram of the refinement of S2 of the present invention;

[0049] Figure 4 It is a schematic diagram of the refinement of S3 of the present invention;

[0050] Figure 5 It is a schematic diagram of the refinement of S4 of the present invention;

[0051] Figure 6 It is a detailed schematic diagram of S5 of the present invention;

[0052] Figure 7 It is a detailed schematic diagram of S6 of the present invention. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0054] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0055] See also Figure 1 The present invention provides a technical solution, a method for early warning and handling of automobile finance post-loan risks, comprising the following steps:

[0056] S1: Based on the borrower's post-loan transaction information, collect and analyze the borrower's bank statements and consumption record information, including the amount, time and frequency of multiple transactions, to generate the borrower's bill data;

[0057] S2: Based on the borrower's bill data, by analyzing the transaction objects, time, and amount of multiple transactions, abnormal transaction information and risky transaction behaviors are identified, and transaction behavior analysis results are generated;

[0058] S3: Based on the transaction behavior analysis results, the value of the vehicle is evaluated by analyzing the market trend, vehicle condition and historical usage data of the borrower's vehicle, and the mortgage asset evaluation results are generated in combination with the borrower's credit record and economic status;

[0059] S4: Using the mortgage asset evaluation results, by analyzing the asset status of multiple borrowers, combined with historical repayment behavior and overdue records, the repayment risk score is calculated to generate risk borrower identification results;

[0060] S5: According to the risk borrower identification results, matching collection notification information for borrowers of multiple risk levels, adjusting the frequency of notification sending, recording the borrower's repayment response behavior, and generating collection response records;

[0061] S6: Use the collection response records, consider the borrower's financial status and capital recovery efficiency, adjust the loan repayment period and overdue interest rate, calculate the repayment amount, and generate post-loan risk processing results.

[0062] The borrower's billing data includes transaction date information, transaction object information, and transaction amount information. The transaction behavior analysis results include abnormal transaction identification records, risky transaction type data, and risky transaction period information. The mortgage asset assessment results include real-time automobile market value assessment data, expected depreciation rate, maintenance and repair records. The risky borrower identification results include credit risk rating, predicted default rate, and borrower financial stress assessment results. The collection response records include response time analysis results, repayment amount change information, and repayment willingness assessment information. The post-loan risk processing results include repayment plan adjustment results, overdue interest rate update results, and expected funds recovery efficiency.

[0063] See also Figure 2 Based on the borrower's post-loan transaction information, the borrower's bank statements and consumption record information are collected and analyzed, including the amount, time and frequency of multiple transactions. The specific steps for generating the borrower's bill data are:

[0064] S101: Based on the borrower's post-loan transaction information, the borrower's bank statements and consumption records are collected, and the transaction date, amount and frequency information are recorded to generate the transaction data collection results. The specific process is as follows;

[0065] In sub-step S101, based on the borrower's post-loan transaction information, the borrower's bank statements and consumption records are collected, including exporting transaction data from the bank system through the API interface, integrating the consumption records of the e-commerce platform, and capturing transaction date, amount and frequency information through data crawler technology. The validity and integrity of the data are verified to ensure that the timestamp and amount of each transaction record are verified. The formula is:

[0066]

[0067] Among them, T ioz represents the adjusted transaction time coefficient, A ioz represents the original transaction amount, k z is the time coefficient adjustment parameter used to standardize the time value in the transaction record, β z It is the offset of the time coefficient adjustment, which is used to standardize the time and amount of transaction records, ensure the consistency and comparability of data, and generate transaction data collection results.

[0068] S102: Based on the transaction data collection results, the data is cleaned and formatted, duplicate records and incorrect inputs are identified, and the specific process of generating consumption record processing results is as follows;

[0069] In sub-step S102, based on the transaction data collection results, the data is cleaned and formatted, including using a custom data cleaning script to identify and remove duplicates and incorrect inputs in the transaction records, locating duplicate transaction records through hash table technology, and automatically correcting obvious input errors using pre-defined rules, and standardizing the transaction data into a unified format. The formula is:

[0070]

[0071] Among them, D jz represents the quality index after data cleaning, X koz Indicates the amount of a single transaction. It is the average value of all transaction amounts, and n0 is the total number of transactions. It is used to evaluate the effect of data cleaning, ensure the consistency and accuracy of the data set, and generate consumption record processing results.

[0072] S103: Based on the consumption record processing result, the transaction data is classified, the total amount and transaction frequency of various transactions are calculated, and the specific process of generating the borrower's bill data is as follows;

[0073] In sub-step S103, based on the consumption record processing results, the transaction data is classified, including using a decision tree algorithm to distinguish transactions of different types, calculating the total amount and transaction frequency of each type of transaction, understanding the consumption behavior pattern and predicting the consumption trend. The formula is:

[0074]

[0075] Among them, C kz represents the average transaction cost of a category, f loz and a loz are the frequency and amount of the l0th transaction respectively, m0 is the total number of transactions in the target category, the consumption intensity of different categories is calculated and compared, and the borrower's bill data is generated.

[0076] See also Figure 3 Based on the borrower's bill data, by analyzing the transaction objects, time, and amount of multiple transactions, abnormal transaction information and risky transaction behaviors are identified. The specific steps for generating transaction behavior analysis results are as follows:

[0077] S201: Based on the borrower's bill data, by analyzing the amounts and transaction times of multiple transactions and comparing them with the preset monitoring threshold, identifying abnormal transaction amounts and transaction times, and generating abnormal transaction comparison results, the specific process is as follows;

[0078] In sub-step S201, based on the borrower's bill data, by analyzing the amount and transaction time of multiple transactions and comparing them with the preset monitoring threshold, the formula used is:

[0079]

[0080] Among them, E i1x Represents the transaction anomaly score, A i1x represents the amount of a single transaction, θ Ax is the preset monitoring threshold of the amount, σ Ax is the standard deviation of the amount, T i1x represents the transaction time, θ Tx is the preset monitoring threshold of time, σ Tx It is the standard deviation over time, identifying transactions that deviate from normal patterns and generating abnormal transaction comparison results.

[0081] S202: Based on the abnormal transaction comparison results, the background information of the abnormal transaction is analyzed, including the payment location and the transaction object, and the risky transactions are classified and recorded. The specific process of generating the risky transaction classification results is as follows;

[0082] In sub-step S202, based on the abnormal transaction comparison results, the background information of the abnormal transaction is analyzed, including the payment location and transaction object. Using the classification results, the frequency and impact of each type of risky transaction are recorded. The formula is:

[0083]

[0084] Among them, R k1x represents the risk score of the target classification, Wp j1x and wo j1x are the risk weight coefficients of the payment location and the transaction object, Pp j1x and Po j1x It is a risk probability score calculated based on the payment location and transaction object. n1 is the total number of transactions in the category. The payment location and transaction object risk scores of each transaction are added up by weighted summation. The classification results of risky transactions are evaluated and recorded to generate risky transaction classification results.

[0085] S203: Based on the risk transaction classification results, by evaluating the impact of various transaction behaviors on the borrower's credit status, using a logistic regression algorithm, calculating the credit impact scores of various transaction behaviors, and generating the transaction behavior analysis results, the specific process is as follows;

[0086] In sub-step S203, based on the risk transaction classification results, by evaluating the impact of various transaction behaviors on the borrower's credit status, a logistic regression algorithm is used to combine the transaction amounts, transaction frequencies, transaction time intervals, the borrower's historical overdue times, and the amount of funds inflow in the borrower's account of multiple transactions to calculate the credit impact score of each behavior, quantify the impact of different transaction behaviors on the borrower's credit status, and generate transaction behavior analysis results.

[0087] Logistic regression algorithm, according to the formula:

[0088]

[0089] Calculate the borrower's default probability, where p is the probability of the borrower's default, b0 is the intercept term, b1 is the weight related to the transaction amount, b2 is the weight related to the transaction frequency, b3 is the weight related to the transaction time interval, b4 is the weight related to the borrower's historical overdue times, b5 is the weight related to the amount of funds flowing into the borrower's account, x1 is the transaction amount, x2 is the transaction frequency, x3 is the transaction time interval, x4 is the borrower's historical overdue times, x5 is the amount of funds flowing into the borrower's account, and e is the base of the natural logarithm.

[0090] Calculate the borrower's default probability, where p is the probability of borrower default, which is used to indicate the possibility of borrower default. b0 is the intercept term, which provides the default value when there is no input variable influence when the model output is provided. b1 is the weight related to the transaction amount, which is used to measure the impact of the amount on the default probability. b2 is the weight related to the transaction frequency, which is used to measure the impact of the transaction frequency on the default probability. b3 is the weight related to the transaction time interval, which is used to measure the impact of the time interval on the default probability. b4 is the weight related to the borrower's historical overdue number, which is used to measure the impact of the borrower's historical overdue behavior on the default probability. b5 is the weight related to the amount of funds inflow into the borrower's account, which is used to measure the impact of the amount of funds inflow on the default probability. x1 is the transaction amount, which is used to indicate the scale of the transaction. x2 is the transaction frequency, which is used to measure the number of active transactions of the borrower. x3 is the time interval of the transaction, which is used to describe the time density of the transaction. x4 is the number of historical overdue payments of the borrower, which is used to measure the borrower's historical credit behavior. x5 is the amount of funds inflow into the borrower's account, which is used to reflect the borrower's financial liquidity. e is the base of the natural logarithm.

[0091] The specific execution process of the formula is as follows:

[0092] Through the borrower's bank statement data, we obtain the parameters of transaction amount, transaction frequency, and transaction time interval. Through the credit report, we obtain the parameters of the borrower's historical overdue times. Through the borrower's account statement, we obtain the parameters of the amount of funds flowing into the borrower's account. Using the logistic regression model, we obtain the parameters b0, b1, b2, b3, b4, and b5. Substituting the parameters into the formula, we calculate the probability of borrower default, which is used to evaluate and manage the borrower's default risk.

[0093] See also Figure 4 Based on the transaction behavior analysis results, the value of the vehicle is evaluated by analyzing the market trend, vehicle condition and historical usage data of the borrower's vehicle. Combined with the borrower's credit record and economic situation, the steps to generate the mortgage asset evaluation results are as follows:

[0094] S301: Based on the transaction behavior analysis results, the borrower's vehicle information is collected, and the basic conditions and historical usage data of the vehicle are recorded, including the vehicle model, age, mileage and maintenance records. The specific process of generating vehicle condition data records is as follows;

[0095] In sub-step S301, based on the transaction behavior analysis results, the basic information and historical usage data of the vehicle are collected through data collection, the model, age, mileage and maintenance record of each vehicle are recorded, and the data format is standardized and the data integrity is checked to ensure data accuracy. The formula is:

[0096]

[0097] Among them, V i2c represents the comprehensive condition score of the vehicle, α c is the weight coefficient of vehicle age, β c is the weight coefficient of mileage, γ c is the weight coefficient of the maintenance times, A i2c is the age of the vehicle, D i2c is the mileage, R k2c is a single maintenance record, m2 is the number of maintenance records, and the overall condition of the vehicle is quantified by comprehensively considering the influence of vehicle age, mileage and maintenance records to generate vehicle condition data records.

[0098] S302: Based on the vehicle condition data record, the sales price and demand trend of various vehicles are analyzed through real-time market data to generate the market trend analysis result. Specifically, the process is as follows;

[0099] In sub-step S302, based on the vehicle condition data record, by connecting to the automobile sales platform and online market data source in real time, using data mining, pattern recognition and data capture algorithms, the sales price and demand trend of different models are collected and analyzed. The formula is:

[0100]

[0101] Among them, S j2c is the market trend score, λ c is the price weight coefficient, μ c is the weight coefficient of demand trend, P j2c is the selling price, T j2c It is the demand trend, evaluating the performance and potential value of different models in the market and generating market trend analysis results.

[0102] S303: Based on the market trend analysis results, the borrower's vehicle value and change trend are evaluated in real time, and the process of generating the mortgage asset evaluation results is specifically as follows:

[0103] In sub-step S303, based on the market trend analysis results, the borrower's vehicle value and change trend are evaluated in real time. By combining the vehicle condition data and market trend data with the borrower's credit record and economic status, financial modeling and credit scoring algorithms are used to analyze the impact of the value of multiple vehicles. The formula is:

[0104]

[0105] Among them, Z k2c is the mortgage asset evaluation result, Vc is the weight coefficient of the vehicle condition score, ξ c is the weight coefficient of the market trend score, ω c is the weight coefficient of credit record, ψ c is the weight coefficient of economic conditions, V k2c is the vehicle condition score, S k2c is the market trend score, C k2c It is the credit record, E k2c It is the economic situation that evaluates the mortgage value of the vehicle by combining the vehicle condition, market performance, borrower credit and economic situation to generate a mortgage asset valuation result.

[0106] See also Figure 5 , using the mortgage asset evaluation results, by analyzing the asset status of multiple borrowers, combined with historical repayment behavior and overdue records, calculating the repayment risk score, the specific steps for generating risk borrower identification results are:

[0107] S401: Using the mortgage asset evaluation results, collect and record the asset status data and historical repayment behavior records of multiple borrowers, including overdue records and repayment frequency information. The specific process of generating asset and behavior data is as follows;

[0108] In sub-step S401, the asset status data and historical repayment records of multiple borrowers are collected and recorded using the mortgage asset evaluation results. The asset information of the borrowers, including the value of real estate and vehicles, is extracted from the database of the financial institution. Combined with the historical repayment behavior data, including overdue records and repayment frequency, the data is timestamped and validated to ensure the accuracy and completeness of the records. The formula is:

[0109]

[0110] Among them, A i3v represents the adjusted asset status index, V i3v represents the original asset value, c v and p v It is an adjustment parameter used to normalize asset values ​​and generate asset and behavior data.

[0111] S402: Based on the asset and behavior data, analyze the borrower's historical repayment behavior and overdue record, evaluate the stability of the repayment behavior and the probability of overdue, and generate the stability analysis results. The specific process is as follows;

[0112] In sub-step S402, based on the asset and behavior data, the borrower's historical repayment behavior and overdue record are analyzed. Through regression analysis, the stability of each borrower's repayment behavior and overdue probability are evaluated. By analyzing the historical data, a prediction model is established to identify the default risk. The formula is:

[0113]

[0114] Among them, S j3v Represents the stability index of repayment behavior, R k3v represents the repayment record of the k3th borrower, is the average of all borrowers’ repayment records, n3 is the total number of borrowers, and the stability of repayment behavior is calculated to generate stability analysis results.

[0115] S403: Based on the stability analysis results, the repayment risk scores of multiple borrowers are calculated, the asset value and credit status are considered, the overdue risk level is evaluated, and the specific process of generating the risk borrower identification results is as follows;

[0116] In sub-step S403, based on the stability analysis results, the repayment risk scores of multiple borrowers are calculated, and the overdue risk level of each borrower is evaluated using a weighted score method taking into account the asset value and credit status of each borrower. The formula is:

[0117]

[0118] Among them, R l3vrepresents the repayment risk score, f m3v and w m3v are the risk factor scores and corresponding weights of the m3th borrower, q3 is the total number of risk factors evaluated, and the risk level of the borrower is evaluated and calculated to generate the risk borrower identification result.

[0119] See also Figure 6 According to the risk borrower identification results, the collection notification information is matched for borrowers of multiple risk levels, the notification sending frequency is adjusted, and the borrower's repayment response behavior is recorded. The specific steps for generating the collection response record are as follows:

[0120] S501: According to the risk borrower identification result, the collection notice content is matched according to the risk levels of multiple borrowers, including the notification information and the sending frequency, and the specific process of generating the collection notice matching result is as follows;

[0121] In sub-step S501, based on the risk borrower identification results, the collection notice content is matched according to the risk levels of multiple borrowers, including the notification information and the sending frequency, and the formula is:

[0122]

[0123] Among them, C i4b Represents the effectiveness index of targeted collection notices, R i4b is the risk level of the borrower, α b , γ b and μ b It is to adjust parameters, optimize the pertinence and efficiency of collection notices, and generate collection notice matching results.

[0124] S502: Based on the collection notice matching result, collection notice information is sent to multiple borrowers, and the repayment behavior of the borrower after receiving the notice is recorded, including the time and amount of repayment. The specific process of generating collection information response data is as follows;

[0125] In sub-step S502, based on the collection notice matching results, collection notice information is sent to multiple borrowers, and the repayment behavior of the borrowers after receiving the notice is recorded, including the repayment time and amount. Through real-time data monitoring, the timeliness and tracking effect of each notification are optimized. The formula is:

[0126]

[0127] Among them, R j4b represents the average repayment amount weighted by the repayment time, t k4b and p k4b They represent the repayment time and amount of the k4th borrower respectively, n4 is the total number of borrowers who received the notice, and the actual effect of the collection notice is analyzed to generate collection information response data.

[0128] S503: Analyze the collection information response data, use the K-means clustering algorithm, analyze the borrower's response speed and repayment completion, evaluate the effects of various collection notice contents, and generate the collection response record. The specific process is as follows;

[0129] In sub-step S503, the collection information response data is analyzed, and the K-means clustering algorithm is used to collect the borrowers' collection response data to obtain the response time, response amount, and repayment behavior parameters, evaluate the effect of the collection notice content, quantify the impact of the collection notice content on the behavior of different borrowers, and generate a collection response record.

[0130] K-means clustering algorithm, according to the formula:

[0131]

[0132] Calculate the clustering effect of the collection strategy, where S is the total internal variance, which is used to measure the overall tightness of the clusters, k is the number of clusters, which is used to specify the number of cluster groups, and C i is the set of data points in the i-th cluster, representing the data points contained in each category, x time is the borrower’s response time data point, used to measure the timeliness of the collection notice, x amount is the borrower’s response amount data point, used to measure the economic impact of the collection notice, x behavior is the borrower's repayment behavior data point, which is used to measure the behavioral effect of the collection notice, μ time is the center point of the time cluster, representing the average response time in each cluster, μ amount is the center point of the amount cluster, representing the average value of the response amount in each cluster, μ behavior is the center point of the behavior cluster, representing the average performance of repayment behavior in each cluster, w1 is the time weight coefficient, adjusting the importance of time in cluster analysis, w2 is the amount weight coefficient, adjusting the importance of amount in cluster analysis, w3 is the behavior weight coefficient, adjusting the importance of behavior in cluster analysis, i is the cluster index, used to distinguish different cluster groups in the summation process, x is a single data point, representing the i A specific instance of , which is used to calculate the distance to the cluster center and evaluate the variance within the cluster.

[0133] The specific execution process of the formula is as follows:

[0134] By collecting the borrower's collection response data, we can get the response time x time 、Response amount x amount 、Repayment behavior x behaviorParameters, based on historical data and predetermined analysis objectives, weight coefficients w1, w2, and w3 are determined using statistical methods. The coefficients are adjusted through gradient descent to ensure the optimization of clustering results. The parameters are substituted into the formula, and the K-means clustering algorithm is used to perform clustering operations and calculate the internal variance S of each cluster to evaluate the impact of different collection notices on borrower behavior, helping financial institutions optimize collection strategies and improve capital recovery efficiency.

[0135] See also Figure 7 , using the collection response record, considering the borrower's financial status and capital recovery efficiency, adjusting the loan repayment period and overdue interest rate and calculating the repayment amount, the steps to generate the post-loan risk processing results are as follows:

[0136] S601: Analyze the collection response records, analyze the repayment amounts and loan interest rates of multiple loan contracts, and compare them with the actual repayment capabilities of multiple borrowers to identify the overdue risks of multiple borrowers. The specific process of generating overdue risk identification data is as follows;

[0137] In sub-step S601, the collection response records are analyzed, and the repayment amounts and loan interest rates of multiple loan contracts are analyzed and compared with the actual repayment capabilities of multiple borrowers to calculate the financial stress index of each borrower and identify the overdue risk. The formula is:

[0138]

[0139] Among them, R i5q represents the overdue risk indicator, which is used to measure the borrower's default possibility. i5q Indicates the borrower's loan interest rate, reflecting the cost of borrowing. i5q represents the borrower's monthly income, which is used as a direct indicator of repayment ability, and q is the overdue risk adjustment coefficient, which is used to adjust the influence of loan interest rate on overdue risk assessment, ε q To prevent a constant with a denominator of zero, overdue risk identification data is generated.

[0140] S602: Based on the overdue risk identification data, taking into account the fund recovery efficiency and the borrower's financial capacity, identifying the repayment period and overdue interest rate that need to be adjusted, and generating the repayment condition adjustment data, the specific process is as follows;

[0141] In sub-step S602, based on the overdue risk identification data, taking into account the fund recovery efficiency and the borrower's financial capacity, the repayment period and overdue interest rate that need to be adjusted are identified, and the fund recovery is optimized by calculating the appropriate repayment period and increased overdue interest rate for each borrower. The formula is:

[0142]

[0143] Among them, P j5q represents the recommended repayment period, which is used to adjust the borrower's repayment plan. j5q Indicates the fund recovery efficiency, which measures the borrower’s past repayment performance. j5q represents the borrower's financial stress index, calculated based on the borrower's income and debt burden, ω q is the cycle adjustment coefficient, which is used to adjust the length of the repayment cycle according to the capital recovery efficiency, θ q It is the power index of the financial pressure adjustment index, which is used to adjust the impact of financial pressure on the repayment period and generate repayment condition adjustment data.

[0144] S603: The specific process of adjusting data according to repayment conditions, adjusting the borrower's repayment contract, monitoring the adjusted repayment behavior and analyzing the impact on repayment efficiency, and generating post-loan risk processing results is as follows;

[0145] In sub-step S603, the data is adjusted according to the repayment conditions, and the borrower's repayment contract is adjusted, including modifying the repayment period and overdue interest rate in the contract, monitoring the adjusted repayment behavior and analyzing the impact on repayment efficiency. The formula is:

[0146]

[0147] Among them, E k5q Represents the improvement index of repayment efficiency, which is used to evaluate the improvement of repayment behavior after contract adjustment. k5q represents the adjusted repayment behavior data, including the frequency and amount of on-time repayments, σ q is the efficiency improvement coefficient, which is used to amplify or reduce the evaluation results of repayment behavior improvement, ρ q It is an adjustment constant used for data stability, ensuring the feasibility of logarithmic calculations, evaluating the impact of adjusting repayment terms on overall financial health, and generating post-loan risk treatment results.

[0148] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A post-loan risk early warning and disposal method for automobile finance, characterized in that: The following steps are involved: Based on the borrower's post-loan transaction information, collect and analyze the borrower's bank statements and consumption record information, including the amount, time and frequency of multiple transactions, to generate the borrower's bill data; Based on the borrower's bill data, by analyzing the transaction objects, time, and amount of multiple transactions, abnormal transaction information and risky transaction behaviors are identified, and transaction behavior analysis results are generated; Based on the transaction behavior analysis results, the value of the vehicle is evaluated by analyzing the market trend, vehicle condition and historical usage data of the borrower's vehicle, and the mortgage asset evaluation results are generated in combination with the borrower's credit record and economic status; Using the mortgage asset evaluation results, by analyzing the asset status of multiple borrowers, combined with historical repayment behaviors and overdue records, a repayment risk score is calculated to generate a risk borrower identification result; According to the risk borrower identification results, matching collection notification information for borrowers of multiple risk levels, adjusting the frequency of sending notifications, recording the borrower's repayment response behavior, and generating collection response records; Using the collection response record, taking into account the borrower's financial status and fund recovery efficiency, adjusting the loan repayment period and overdue interest rate and calculating the repayment amount, and generating a post-loan risk processing result; Among them, based on the borrower's post-loan transaction information, the borrower's bank statements and consumption record information are collected and analyzed, including the amount, time and frequency of multiple transactions. The specific steps for generating the borrower's bill data are: S101: Based on the borrower's post-loan transaction information, collect the borrower's bank statements and consumption records, record the transaction date, amount and frequency information, and generate transaction data collection results; S102: Based on the transaction data collection results, clean and format the data, identify duplicate records and incorrect inputs, and generate consumption record processing results; S103: Based on the consumption record processing results, the transaction data is classified, the total amount and transaction frequency of various transactions are calculated, and the borrower's bill data is generated. In the S103 sub-step, the average transaction cost of each category is calculated by the following formula to obtain the corresponding consumption intensity: Among them, C kz represents the average transaction cost of a category, f loz and a loz are the frequency and amount of the l0th transaction, respectively, and m0 is the total number of transactions in the target category; Among them, according to the risk borrower identification result, matching the collection notification information for borrowers of multiple risk levels, adjusting the sending frequency of the notification, recording the repayment response behavior of the borrower, and generating the collection response record are specifically as follows: According to the risk borrower identification result, matching the collection notice content according to the risk levels of multiple borrowers, including the notification information and the sending frequency, to generate a collection notice matching result; Based on the collection notice matching result, collection notice information is sent to multiple borrowers, and the repayment behavior of the borrowers after receiving the notice is recorded, including the time and amount of repayment, to generate collection information response data; The collection information response data is analyzed, and the K-means clustering algorithm is used to analyze the borrower's response speed and repayment completion, evaluate the effects of various collection notice contents, and generate collection response records; the K-means clustering algorithm is based on the formula: Calculate the clustering effect of the collection strategy, where S is the total internal variance, k is the number of clusters, and C i is the set of data points in the i-th cluster, x time is the borrower’s response time data point, x amount is the borrower’s response amount data point, x behavior is the borrower’s repayment behavior data point, μ time is the center point of the time cluster, μ amount is the center point of the amount cluster, μ behavior is the center point of the behavior cluster, w1 is the time weight coefficient, w2 is the amount weight coefficient, w3 is the behavior weight coefficient, i is the cluster index, and x is a single data point.

2. The automobile finance post-loan risk early warning and disposal method according to claim 1, characterized in that: The borrower billing data includes transaction date information, transaction object information, and transaction amount information; the transaction behavior analysis results include abnormal transaction identification records, risky transaction type data, and risky transaction period information; the mortgage asset assessment results include real-time automobile market value assessment data, expected depreciation rate, maintenance and repair records; the risky borrower identification results include credit risk rating, predicted default rate, and borrower financial stress assessment results; the collection response records include response time analysis results, repayment amount change information, and repayment willingness assessment information; the post-loan risk processing results include repayment plan adjustment results, overdue interest rate update results, and expected funds recovery efficiency.

3. The automobile finance post-loan risk early warning and disposal method according to claim 1, characterized in that: Based on the borrower's bill data, by analyzing the transaction objects, time, and amount of multiple transactions, identifying abnormal transaction information and risky transaction behaviors, the steps of generating transaction behavior analysis results are specifically as follows: Based on the borrower's bill data, by analyzing the amounts and transaction times of multiple transactions and comparing them with preset monitoring thresholds, abnormal transaction amounts and transaction times are identified, and abnormal transaction comparison results are generated; Based on the abnormal transaction comparison results, analyzing the background information of the abnormal transaction, including the payment location and the transaction object, classifying and recording the risky transactions, and generating risky transaction classification results; According to the risk transaction classification results, by evaluating the impact of various transaction behaviors on the borrower's credit status, a logistic regression algorithm is used to calculate the credit impact scores of various transaction behaviors to generate transaction behavior analysis results.

4. The automobile finance post-loan risk early warning and disposal method according to claim 3 is characterized in that: The logistic regression algorithm is based on the formula: Calculate the borrower's default probability, where p is the probability of the borrower's default, b0 is the intercept term, b1 is the weight related to the transaction amount, b2 is the weight related to the transaction frequency, b3 is the weight related to the transaction time interval, b4 is the weight related to the borrower's historical overdue times, b5 is the weight related to the amount of funds flowing into the borrower's account, x1 is the transaction amount, x2 is the transaction frequency, x3 is the transaction time interval, x4 is the borrower's historical overdue times, x5 is the amount of funds flowing into the borrower's account, and e is the base of the natural logarithm.

5. The automobile finance post-loan risk early warning and disposal method according to claim 1, characterized in that: Based on the transaction behavior analysis results, the value of the vehicle is evaluated by analyzing the market trend, vehicle condition and historical usage data of the borrower's vehicle. Combined with the borrower's credit record and economic situation, the steps for generating the mortgage asset evaluation results are as follows: Based on the transaction behavior analysis results, the borrower's vehicle information is collected, the basic conditions and historical usage data of the vehicle are recorded, including the vehicle model, age, mileage and maintenance records, and a vehicle condition data record is generated; Based on the vehicle condition data records, analyzing the sales prices and demand trends of various vehicles through real-time market data to generate market trend analysis results; Based on the market trend analysis results, the borrower's vehicle value and change trend are evaluated in real time, and the mortgage asset evaluation results are generated in combination with the borrower's credit record and economic status.

6. The automobile finance post-loan risk early warning and disposal method according to claim 1, characterized in that: Using the mortgage asset evaluation results, by analyzing the asset status of multiple borrowers, combined with historical repayment behavior and overdue records, the repayment risk score is calculated, and the steps for generating risk borrower identification results are as follows: Using the mortgage asset evaluation results, collect and record asset status data and historical repayment behavior records of multiple borrowers, including overdue records and repayment frequency information, to generate asset and behavior data; Based on the asset and behavior data, analyze the borrower's historical repayment behavior and overdue records, evaluate the stability of repayment behavior and overdue probability, and generate stability analysis results; Based on the stability analysis results, the repayment risk scores of multiple borrowers are calculated, the asset value and credit status are considered, the overdue risk level is evaluated, and the risk borrower identification results are generated.

7. The automobile finance post-loan risk early warning and disposal method according to claim 1, characterized in that: Using the collection response record, taking into account the borrower's financial status and capital recovery efficiency, adjusting the loan repayment period and overdue interest rate and calculating the repayment amount, the steps for generating the post-loan risk processing result are as follows: Analyze the collection response records, analyze the repayment amounts and loan interest rates of multiple loan contracts, and compare them with the actual repayment capabilities of multiple borrowers to identify the overdue risks of multiple borrowers and generate overdue risk identification data; Based on the overdue risk identification data, taking into account the fund recovery efficiency and the borrower's financial capacity, identifying the repayment period and overdue interest rate that need to be adjusted, and generating repayment condition adjustment data; Adjust the data according to the repayment conditions, adjust the borrower's repayment contract, monitor the adjusted repayment behavior and analyze the impact on repayment efficiency, and generate post-loan risk management results.

Images