A financial product matching processing method based on customer big data and vehicle information

Abstract

The present invention discloses a financial product matching and processing method based on customer big data and vehicle information, which relates to the field of financial technology, including collecting customer data and vehicle data, preprocessing the collected data; extracting characteristic values ​​of the collected data after preprocessing, and performing customer behavior risk assessment based on the extracted characteristic values ​​of the collected data to obtain a risk score of the customer behavior; determining the customer risk range according to the customer behavior risk score, and recommending the bank's vehicle loan products to the customer through a visual interface. By combining customer big data with vehicle information, the accuracy of financial product matching is comprehensively improved, and by collecting multi-source data, the comprehensiveness of the data is ensured, and a customer behavior risk assessment formula is used to accurately calculate the risk score of the customer behavior and classify it according to the risk preference and business needs of the financial institution. Bank vehicle loan products suitable for their risk type are recommended to customers through a visual interface to meet the personalized needs of customers with different risks.

Description

Technical Field

[0001] The present invention relates to the field of financial technology, and in particular to a financial product matching processing method based on customer big data and vehicle information. Background Art

[0002] With the rapid development of big data technology and artificial intelligence, product recommendation and risk assessment technologies in the financial field are also constantly evolving. Traditional financial product matching methods mainly rely on manual experience and fixed rule systems. This method is inefficient and difficult to accurately reflect customers' real needs and risk conditions. In recent years, more and more financial institutions have begun to adopt big data technology to more accurately conduct customer risk assessment and product recommendations through the analysis of massive customer data and related behavioral data. This technological trend has not only improved the service efficiency of financial institutions, but also significantly improved customer satisfaction.

[0003] Although existing related technologies have achieved personalized recommendations for financial products to a certain extent, there are still many shortcomings. Traditional financial product matching systems mostly rely on historical data and single-dimensional analysis, lack comprehensive evaluation of multi-source heterogeneous data, and have low accuracy and timeliness in customer behavior risk assessment, making it difficult to dynamically reflect changes in customer risks. Existing technologies have limited use of vehicle information and fail to fully tap the potential correlation between vehicle data and customer financial needs. Summary of the invention

[0004] In view of the problems existing in the above-mentioned existing financial product matching processing method based on customer big data and vehicle information, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is that the existing financial product matching technology data analysis is incomplete, the risk assessment accuracy is insufficient, and the personalized recommendation effect is poor.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a financial product matching processing method based on customer big data and vehicle information, which comprises:

[0007] Collect customer data and vehicle data, and pre-process the collected data;

[0008] Extracting characteristic values ​​of the preprocessed collected data, and performing customer behavior risk assessment based on the extracted characteristic values ​​of the collected data to obtain a risk score of the customer behavior;

[0009] Determine the customer risk range based on the customer behavior risk score and recommend the bank's vehicle loan products to the customer through a visual interface;

[0010] Customer data and vehicle data are stored in encrypted form.

[0011] As a preferred solution of the financial product matching processing method based on customer big data and vehicle information described in the present invention, wherein: the collection of customer data and vehicle data refers to determining the source of customer and customer vehicle data, developing an API interface, and deploying the big data collection tool Apache Flume to collect customer and customer vehicle data in real time. The collected data includes customer information, vehicle information, customer financial behavior data, vehicle usage behavior data, insurance claims and driving violation records.

[0012] As a preferred solution of the financial product matching and processing method based on customer big data and vehicle information described in the present invention, the preprocessing of the collected data refers to preprocessing the collected data, including deleting duplicate data, processing missing values, identifying and correcting outliers, formatting the collected data, and normalizing the collected data.

[0013] As a preferred solution of the financial product matching processing method based on customer big data and vehicle information of the present invention, wherein: the extracting of pre-processed data features refers to integrating the customer and customer vehicle data sets using customer names and license plate numbers according to customer information and vehicle information, and using Python's Pandas library to extract data features from the customer and customer vehicle data sets, including customer financial behavior features, vehicle use behavior features, and insurance claims and driving violation record features, to form a customer feature matrix X:

[0014]

[0015] Among them, X pn is the nth feature of the pth customer and customer vehicle dataset;

[0016] Based on the historical data of customers and their vehicles, the driving violation records in the historical data of customers and their vehicles are defined as a binary target variable. If the customer has a driving violation record in the past year, the binary target variable is defined as 1, otherwise it is 0;

[0017] Composition of binary target variable matrix Y:

[0018]

[0019] Among them, y p represents the binary target variable of the p-th customer and customer vehicle dataset;

[0020] Calculate the point biserial correlation coefficient r to determine the correlation between the features in the customer feature matrix and the binary target variable:

[0021]

[0022] in, The feature X is 1 for the binary target variable y f The mean of The feature X is 0 for the binary target variable y g The mean value, S x For feature X pn The standard deviation, p 1 is the number of customers and customer vehicle datasets where the target variable y is 1, p 0 is the number of customer and customer vehicle datasets where the target variable y is 0, and p is the total number of customer and vehicle datasets;

[0023] Set feature threshold r th , if |r|≤r th , then the feature is not extracted. If |r|>r th , then extract the feature.

[0024] As a preferred solution of the financial product matching processing method based on customer big data and vehicle information of the present invention, wherein: the risk score of customer behavior obtained by performing customer behavior risk assessment based on the extracted features includes:

[0025] Customer financial behavior characteristics include credit card usage frequency characteristic value, credit card limit value characteristic value, bank account balance change characteristic value and bank account average balance characteristic value;

[0026] Substituting the customer's financial behavior characteristics into the financial behavior characteristic function A(t), the formula is:

[0027]

[0028] Among them, ω 1 and ω 2 To adjust the weight coefficients of the influence of different financial behaviors, CU represents the characteristic value of the customer's credit card usage frequency at time t, CL represents the characteristic value of the customer's credit card limit value, BA represents the characteristic value of the customer's bank account balance change at time t, and AB represents the characteristic value of the average balance of the customer's bank account;

[0029] Vehicle usage behavior characteristics include average monthly mileage characteristic value, vehicle age characteristic value, and vehicle usage frequency characteristic value;

[0030] Substituting the vehicle usage behavior characteristics into the vehicle usage behavior characteristic function M(t), the formula is:

[0031]

[0032] Among them, φ 1 and φ 2To adjust the weight coefficients of the influence of different vehicle usage behaviors, ML represents the characteristic value of the customer's average monthly mileage at time t, VA represents the characteristic value of the vehicle age, and UF represents the characteristic value of the customer's vehicle usage frequency at time t;

[0033] The weight coefficient for calculating historical behavior data is:

[0034]

[0035] Among them, α 1 is the weight coefficient of historical behavior data, ω 1 and ω 2 To adjust the weight coefficients of the influence of different financial behaviors, φ 1 and φ 2 To adjust the weight coefficients of the influence of different vehicle usage behaviors;

[0036] According to the customer's financial behavior characteristics and vehicle usage behavior characteristics, the cumulative impact value Q of the customer's historical behavior characteristics is calculated. The formula is:

[0037]

[0038] In the formula, Q is the cumulative impact value of the customer's historical behavior characteristics, a is the start time of the historical behavior, b is the end time of the historical behavior, α 1 is the weight coefficient of historical behavior data, λ is the time decay factor, A(t) represents the customer's financial behavior characteristic function at time t, M(t) represents the customer's vehicle usage behavior characteristic function at time t, dt is a small increment in time, representing the integral variable;

[0039] The insurance claim and driving violation record features include the total number of insurance claims feature values ​​and the total number of driving violations feature values;

[0040] The total number of insurance claims and the total number of driving violations are used as risk factors, and the value of the risk factor is the total number. The comprehensive impact value E of the customer risk factor is calculated using the formula:

[0041]

[0042] Where E is the comprehensive impact value of customer risk factors, m is the number of risk factors, γ j is the weight of the jth risk factor, V j is the value of the j-th risk factor;

[0043] Based on the customer's financial behavior characteristics, vehicle use behavior characteristics, insurance claims and driving violation record characteristics, the combined effect value T of the characteristics is calculated using the formula:

[0044]

[0045] Wherein, T is the combined effect value of each customer feature, u is the number of features, and β i is the weight of the i-th feature, and C io is the o-th feature value in the i-th feature;

[0046] Construct a customer behavior risk assessment formula to obtain a customer behavior risk score R. The formula is:

[0047]

[0048] Wherein, R is the customer risk score, k is the calibration coefficient, and s x is the feature standard deviation.

[0049] As a preferred solution of the financial product matching and processing method based on customer big data and vehicle information according to the present invention, wherein: determining the customer risk interval according to the customer behavior risk score means obtaining the customer behavior risk score R through the customer behavior risk assessment formula;

[0050] According to the risk preference and business requirements of the financial institution, map the calculated customer behavior risk score R to the set value range [0, 1] through normalization;

[0051] Set judgment thresholds R1 and R2, and compare the obtained customer behavior risk score R with R1 and R2 to judge the risk type of the customer:

[0052] If R < R2, it indicates that this type of customer is a low-risk customer;

[0053] If R2 ≤ R < R1, it indicates that this type of customer is a medium-risk customer;

[0054] If R ≥ R1, it indicates that this type of customer is a high-risk customer.

[0055] As a preferred solution of the financial product matching and processing method based on customer big data and vehicle information according to the present invention, wherein: recommending the vehicle loan products of the bank to the customer through the visual interface means constructing a database containing all the vehicle loan products of the banks that can be recommended, and each vehicle loan product corresponds to a different risk interval,

[0056] If it is a low-risk customer, recommend the low-interest vehicle loan and fixed-rate loan of large state-owned banks;

[0057] If it is a medium-risk customer, recommend the medium-interest vehicle loan and floating-rate vehicle loan of joint-stock commercial banks;

[0058] If it is a high-risk customer, recommend the credit repair vehicle loan of local banks;

[0059] Select a user-based collaborative filtering algorithm, and use the similarity formula to find customers with similar behaviors to the current customer. Based on the product preferences of similar customers, recommend vehicle loan products preferred by similar customers through a visual interface.

[0060] As a preferred solution of the financial product matching processing method based on customer big data and vehicle information described in the present invention, the encrypted storage of customer and vehicle data refers to selecting the symmetric encryption algorithm AES for data encryption, using a secure key algorithm to generate a 256-bit encryption key, using a secure key management service to store and manage the encryption key, and storing the encrypted data in a distributed file system.

[0061] A computer device includes: a memory and a processor; the memory stores a computer program, and the processor implements the steps of the above-mentioned financial product matching processing method based on customer big data and vehicle information when executing the computer program.

[0062] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned financial product matching processing method based on customer big data and vehicle information.

[0063] The beneficial effects of the present invention are as follows: by combining customer big data with vehicle information, the accuracy of financial product matching is comprehensively improved; by collecting multi-source data, the comprehensiveness of data is ensured; by adopting the customer behavior risk assessment formula, the risk score of customer behavior is accurately calculated and classified according to the risk preference and business needs of the financial institution; through a visual interface, bank vehicle loan products suitable for their risk type are recommended to customers, thereby meeting the personalized needs of customers with different risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0065] Figure 1 The figure is a flowchart of a method for matching financial products based on customer big data and vehicle information.

[0066] Figure 2 Schematic diagram of the structure of customer behavior risk assessment. DETAILED DESCRIPTION

[0067] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0068] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0069] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments.

[0070] Example 1

[0071] Reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and provides a financial product matching processing method based on customer big data and vehicle information. The financial product matching processing method based on customer big data and vehicle information includes:

[0072] S1. Collect customer and customer vehicle data and pre-process the collected data;

[0073] Specifically, collecting customer and vehicle data means determining the sources of customer and customer vehicle data, developing API interfaces, and deploying the big data collection tool Apache Flume to collect customer and vehicle data in real time. Apache Flume uses a distributed architecture and supports the collection of multiple data streams. The collected data includes customer information, vehicle information, customer financial behavior data, vehicle usage behavior data, insurance claims, and driving violation records.

[0074] By determining the source of customer and vehicle data, developing API interfaces and deploying the big data collection tool Apache Flume, real-time collection of customer information, vehicle information, customer financial behavior data, vehicle usage behavior data, insurance claims and driving violation records is achieved. This process ensures the comprehensiveness and timeliness of data collection, and can provide a multi-dimensional, high-quality data foundation, providing a rich source of information for subsequent analysis and evaluation. Through real-time data collection, the system can more accurately capture the dynamic changes of customers and vehicles, improve the timeliness of data, and thus enhance the accuracy and reliability of risk assessment and financial product matching.

[0075] Furthermore, preprocessing the collected data refers to preprocessing the collected data, including deleting duplicate data, processing missing values, identifying and correcting outliers, formatting the collected data, and normalizing the collected data.

[0076] Preprocessing the collected data, ensuring the uniqueness and accuracy of the data by deleting duplicate data, handling missing values ​​and identifying and correcting outliers, improves the integrity and reliability of the data, formats data of different data types, and enhances the model's ability to handle non-numerical features. The maximum and minimum normalization method enables the data to be processed under the same dimension, avoiding model deviations caused by different feature value ranges.

[0077] S2, extracting the feature values ​​of the preprocessed data, and performing a customer behavior risk assessment based on the extracted feature values ​​to obtain a risk score of the customer behavior;

[0078] Specifically, extracting preprocessed data features means integrating the customer and customer vehicle data sets using customer names and license plate numbers based on customer information and vehicle information, and using Python's Pandas library to extract data features from the customer and customer vehicle data sets, including customer financial behavior features, vehicle usage behavior features, and insurance claims and driving violation record features, to form a customer feature matrix X:

[0079]

[0080] Among them, X pn is the nth feature of the pth customer and customer vehicle dataset;

[0081] Based on the historical data of customers and their vehicles, the driving violation records in the historical data of customers and their vehicles are defined as a binary target variable. If the customer has a driving violation record in the past year, the binary target variable is defined as 1, otherwise it is 0;

[0082] Composition of binary target variable matrix Y:

[0083]

[0084] Among them, y p represents the binary target variable of the p-th customer and customer vehicle data set. According to the binary target variable, the feature X pn For classification, when the binary target variable y is 1, it is classified as feature X f , when the binary target variable y is 0, it is classified as feature X g ;

[0085] Calculate the point biserial correlation coefficient r to determine the correlation between the features in the customer feature matrix and the binary target variable:

[0086]

[0087] in, The feature X is 1 for the binary target variable yf The mean of The feature X is 0 for the binary target variable y g The mean value, S x For feature X pn The standard deviation, p 1 is the number of customers and customer vehicle datasets where the target variable y is 1, p 0 is the number of customer and customer vehicle datasets where the target variable y is 0, and p is the total number of customer and vehicle datasets;

[0088] Set feature threshold r th , if |r|≤r th , then the feature is not extracted. If |r|>r th , then extract the feature.

[0089] By using Python's Pandas library to extract all features from a unified data set and form a feature matrix X, the system can efficiently organize and manage a large number of customer and vehicle data samples. By collaborating with the business team and domain experts, the risk customer types are defined and risk feature standards are set based on historical data. Binary target variables are created to achieve accurate classification of customer risks. The point-by-serial correlation coefficient r is further calculated to determine the correlation between the features and the binary target variables, making the feature screening process more scientific.

[0090] Furthermore, based on the extracted features, the customer behavior risk assessment is performed to obtain the customer behavior risk score including:

[0091] Customer financial behavior characteristics include credit card usage frequency characteristic value, credit card limit value characteristic value, bank account balance change characteristic value and bank account average balance characteristic value;

[0092] Substituting the customer's financial behavior characteristics into the financial behavior characteristic function A(t), the formula is:

[0093]

[0094] Among them, ω 1 and ω 2 To adjust the weight coefficients of the influence of different financial behaviors, CU represents the characteristic value of the customer's credit card usage frequency at time t, CL represents the characteristic value of the customer's credit card limit, BA represents the characteristic value of the customer's bank account balance change at time t, and AB represents the characteristic value of the average balance of the customer's bank account;

[0095] Vehicle usage behavior characteristics include average monthly mileage characteristic value, vehicle age characteristic value, and vehicle usage frequency characteristic value;

[0096] Substituting the vehicle usage behavior characteristics into the vehicle usage behavior characteristic function M(t), the formula is:

[0097]

[0098] Among them, φ 1 and φ 2 To adjust the weight coefficients of the influence of different vehicle usage behaviors, ML represents the characteristic value of the customer's average monthly mileage at time t, VA represents the characteristic value of the vehicle age, and UF represents the characteristic value of the customer's vehicle usage frequency at time t;

[0099] The weight coefficient for calculating historical behavior data is:

[0100]

[0101] Among them, α 1 is the weight coefficient of historical behavior data, ω 1 and ω 2 To adjust the weight coefficients of the influence of different financial behaviors, φ 1 and φ 2 To adjust the weight coefficients of the influence of different vehicle usage behaviors;

[0102] According to the customer's financial behavior characteristics and vehicle usage behavior characteristics, the cumulative impact value Q of the customer's historical behavior characteristics is calculated. The formula is:

[0103]

[0104] In the formula, Q is the cumulative impact value of the customer's historical behavior characteristics, a is the start time of the historical behavior, b is the end time of the historical behavior, α 1 is the weight coefficient of historical behavior data, λ is the time decay factor, A(t) represents the customer's financial behavior characteristic function at time t, M(t) represents the customer's vehicle usage behavior characteristic function at time t, dt is a small increment in time, representing the integral variable;

[0105] The insurance claim and driving violation record features include the total number of insurance claims feature values ​​and the total number of driving violations feature values;

[0106] The total number of insurance claims and the total number of driving violations are used as risk factors, and the value of the risk factor is the total number. The comprehensive impact value E of the customer risk factor is calculated using the formula:

[0107]

[0108] Where E is the comprehensive impact value of customer risk factors, m is the number of risk factors, γ j is the weight of the jth risk factor, V j is the value of the j-th risk factor;

[0109] Based on the customer's financial behavior characteristics, vehicle use behavior characteristics, insurance claims and driving violation record characteristics, the combined effect value T of the characteristics is calculated using the formula:

[0110]

[0111] In the formula, T is the combined effect value of each customer feature, u is the number of features, β i is the weight of the i-th feature, C io is the oth eigenvalue in the i-th feature;

[0112] Construct a customer behavior risk assessment formula to obtain the customer behavior risk score R, the formula is:

[0113]

[0114] In the formula, R is the customer risk score, k is the calibration coefficient, and s x is the characteristic standard deviation.

[0115] Through the above steps, the present invention uses the extracted customer financial behavior characteristics and vehicle usage behavior characteristics to calculate the cumulative impact value Q of the customer's historical behavior characteristics, and dynamically adjusts the weight coefficient through regression analysis and the insights of field experts to ensure that the importance of the characteristics is reflected. Financial behavior characteristics such as credit card utilization rate, limit, bank account balance change and average balance, and vehicle usage behavior characteristics such as average monthly mileage, vehicle age and usage frequency are integrated into the customer behavior risk assessment formula, providing a comprehensive risk assessment basis, and determining the weight vector by minimizing the mean square error method to construct the customer feature combination effect value T. Risk factors are extracted from insurance claims and driving violation records, and their weights are determined using a machine learning model to calculate the comprehensive impact value E of customer risk factors. Through the constructed customer behavior risk assessment formula, a customer behavior risk score is obtained.

[0116] S3. Determine the customer risk range based on the customer behavior risk score and recommend the bank's vehicle loan products to the customer through a visual interface;

[0117] Specifically, determining the customer risk range according to the customer behavior risk score means obtaining the customer behavior risk score R through a customer behavior risk assessment formula;

[0118] According to the risk preferences and business needs of financial institutions, the calculated customer behavior risk score R is mapped to the set value range [0, 1] through normalization;

[0119] Set judgment thresholds R1 and R2, determine the thresholds through cross-validation and A / B testing, adjust and optimize the thresholds according to the business of the financial institution, and compare the obtained customer behavior risk score R with R1 and R2 to determine the customer's risk type:

[0120] If R < R2, it indicates that this type of customer is a low-risk customer;

[0121] If R2 ≤ R < R1, it indicates that this type of customer is a medium-risk customer;

[0122] If R ≥ R1, it indicates that this type of customer is a high-risk customer.

[0123] By calculating the customer behavior risk score through the customer behavior risk assessment formula and mapping it to the value range after normalization, the present invention realizes the precise quantification and classification of customer risks. By setting the judgment thresholds R1 and R2 and mapping the calculated customer behavior risk score R to the set value range through normalization, customers are effectively divided into three risk intervals: low risk, medium risk, and high risk. This process not only improves the scientificity and accuracy of risk assessment but also enables financial institutions to recommend appropriate financial products according to the risk levels of customers, thereby optimizing resource allocation and enhancing customer satisfaction and loyalty.

[0124] Furthermore, recommending the bank's vehicle loan products to customers through the visual interface means constructing a database containing all recommendable bank vehicle loan products, and each vehicle loan product corresponds to a different risk interval.

[0125] If it is a low-risk customer, recommend the low-interest vehicle loans and fixed-rate loans of large state-owned banks;

[0126] The borrowing behaviors of low-risk customers are stable. The low interest rate can effectively reduce the repayment pressure of customers and improve customer satisfaction. The fixed interest rate provides a stable repayment environment, enhancing customers' sense of trust and security;

[0127] If it is a medium-risk customer, recommend the medium-term vehicle loans and floating-rate vehicle loans of joint-stock commercial banks;

[0128] Medium-risk customers have certain borrowing credit problems. By giving a certain loan term, it can not only reduce the risks of banks but also meet the capital needs of customers. In addition, floating-rate vehicle loans can adjust the interest rate according to market conditions, providing flexibility for medium-risk customers;

[0129] If it is a high-risk customer, recommend the credit repair vehicle loans of local banks;

[0130] The borrowing risks of high-risk users are relatively high. Recommending the credit repair vehicle loans of local banks not only provides loan funds but also comes with credit repair services to help high-risk customers improve their credit records;

[0131] Select a user-based collaborative filtering algorithm, and use the similarity formula to find customers with similar behaviors to the current customer. Based on the product preferences of similar customers, recommend vehicle loan products that may be of interest to similar customers through a visual interface.

[0132] By building a database containing all recommendable vehicle loan products and using a visual interface to recommend corresponding products based on the customer's risk type, such as recommending low-interest and fixed-rate vehicle loans to low-risk customers, medium-term and floating-rate vehicle loans to medium-risk customers, and credit repair vehicle loans to high-risk customers, it can not only effectively reduce customer repayment pressure and improve satisfaction and trust, but also meet the funding needs and flexibility of customers with different risks. By utilizing a user-based collaborative filtering algorithm, recommendations are further optimized based on the product preferences of similar customers, thereby improving the matching degree of vehicle loan products and customer acceptance, thereby achieving a win-win situation for banks and customers.

[0133] S4, encrypt and store customer and vehicle data;

[0134] Specifically, encrypted storage of customer and vehicle data means selecting the symmetric encryption algorithm AES for data encryption, using a secure key algorithm to generate a 256-bit encryption key, using a secure key management service to store and manage the encryption key, and storing the encrypted data in a distributed file system.

[0135] Customer and vehicle data are encrypted and stored, and the symmetric encryption algorithm AES is selected for data encryption. A 256-bit encryption key is generated using a secure key algorithm to ensure the high security of the encryption process. The encryption keys are stored and managed through secure key management services to effectively prevent key leakage and unauthorized access, thereby ensuring the security of data during storage and transmission. The encrypted data is stored in a distributed file system, which further improves the reliability and scalability of data storage and ensures data security in a large-scale data environment.

[0136] Example 2

[0137] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0138] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0139] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0140] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

Claims

1. A financial product matching processing method based on customer big data and vehicle information, characterized by: include, Collect customer data and vehicle data, and pre-process the collected data; Extracting the features of the preprocessed collected data, and conducting a customer behavior risk assessment based on the extracted collected data features to obtain a risk score for the customer behavior; Determine the customer risk range based on the customer behavior risk score and recommend the bank's vehicle loan products to the customer through a visual interface; Encrypted storage of customer data and vehicle data; Extracting the preprocessed data features refers to integrating the customer and customer vehicle data sets using customer names and license plate numbers based on customer information and vehicle information, and using Python's Pandas library to extract data features from the customer and customer vehicle data sets, including customer financial behavior features, vehicle usage behavior features, and insurance claims and driving violation record features, to form a customer feature matrix X: Among them, X pn is the nth feature of the pth customer and customer vehicle dataset; Based on the historical data of customers and their vehicles, the driving violation records in the historical data of customers and their vehicles are defined as a binary target variable. If the customer has a driving violation record in the past year, the binary target variable is defined as 1, otherwise it is 0; Composition of binary target variable matrix Y: Among them, y p represents the binary target variable of the p-th customer and customer vehicle data set. According to the binary target variable, the feature X pn For classification, when the binary target variable y is 1, it is classified as feature X f , when the binary target variable y is 0, it is classified as feature X g ; Calculate the point biserial correlation coefficient r to determine the correlation between the features in the customer feature matrix and the binary target variable: in, The feature X is 1 for the binary target variable y f The mean of The feature X is 0 for the binary target variable y g The mean value, S x For feature X pn The standard deviation of , p1 is the number of customer and customer vehicle data sets where the target variable y is 1, p0 is the number of customer and customer vehicle data sets where the target variable y is 0, and p is the total number of customer and vehicle data sets; Set feature threshold r th , if |r|≤r th , then the feature is not extracted. If |r|>r th , then extract the feature; The risk score of the customer behavior obtained by performing the customer behavior risk assessment based on the extracted features includes: Customer financial behavior characteristics include credit card usage frequency characteristic value, credit card limit value characteristic value, bank account balance change characteristic value and bank account average balance characteristic value; Substituting the customer's financial behavior characteristics into the financial behavior characteristic function A(t), the formula is: Among them, ω1 and ω2 are weight coefficients for adjusting the influence of different financial behaviors, CU represents the characteristic value of the customer's credit card usage frequency at time t, CL represents the characteristic value of the customer's credit card limit value, BA represents the characteristic value of the customer's bank account balance change at time t, and AB represents the characteristic value of the average balance of the customer's bank account; Vehicle usage behavior characteristics include average monthly mileage characteristic value, vehicle age characteristic value, and vehicle usage frequency characteristic value; Substituting the vehicle usage behavior characteristics into the vehicle usage behavior characteristic function M(t), the formula is: Among them, φ1 and φ2 are weight coefficients for adjusting the influence of different vehicle usage behaviors, ML represents the average monthly mileage characteristic value of the customer at time t, VA represents the characteristic value of vehicle age, and UF represents the characteristic value of the customer's vehicle usage frequency at time t; The weight coefficient for calculating historical behavior data is: Among them, α1 is the weight coefficient of historical behavior data, ω1 and ω2 are the weight coefficients for adjusting the influence of different financial behaviors, and φ1 and φ2 are the weight coefficients for adjusting the influence of different vehicle usage behaviors; According to the customer's financial behavior characteristics and vehicle usage behavior characteristics, the cumulative impact value Q of the customer's historical behavior characteristics is calculated. The formula is: Wherein, Q is the cumulative impact value of the historical behavior characteristics of the customer, a is the start time of the historical behavior, b is the end time of the historical behavior, α1 is the weight coefficient of the historical behavior data, λ is the time decay factor, A(t) represents the financial behavior characteristic function of the customer at time t, M(t) represents the vehicle usage behavior characteristic function of the customer at time t, dt is the infinitesimal increment in time, representing the variable of integration; The insurance claim and driving violation record characteristics include the total number of insurance claim characteristic values and the total number of driving violation characteristic values; Taking the total number of insurance claims and the total number of driving violations as risk factors respectively, and the value of the risk factor is the total number, calculate the comprehensive impact value E of the customer risk factor, and the formula is: Where E is the comprehensive impact value of customer risk factors, m is the number of risk factors, γ j is the weight of the jth risk factor, V j is the value of the j-th risk factor; According to the customer's financial behavior characteristics, vehicle usage behavior characteristics, and insurance claim and driving violation record characteristics, calculate the combined effect value T of the characteristics, and the formula is: In the formula, T is the combined effect value of each customer feature, u is the number of features, β i is the weight of the i-th feature, C io is the oth eigenvalue in the i-th feature; Construct a customer behavior risk assessment formula to obtain the customer behavior risk score R, and the formula is: In the formula, R is the customer risk score, k is the calibration coefficient, and s x is the characteristic standard deviation.

2. The financial product matching processing method based on customer big data and vehicle information as claimed in claim 1, characterized in that: The collection of customer data and vehicle data refers to determining the sources of customer and customer vehicle data, developing API interfaces, and deploying the big data collection tool Apache Flume to collect customer and customer vehicle data in real time. The collected data includes customer information, vehicle information, customer financial behavior data, vehicle usage behavior data, insurance claims and driving violation records.

3. The financial product matching processing method based on customer big data and vehicle information as claimed in claim 1, characterized in that: The preprocessing of the collected data refers to preprocessing the collected data, including deleting duplicate data, handling missing values, identifying and correcting outliers, formatting the collected data, and normalizing the collected data.

4. The financial product matching processing method based on customer big data and vehicle information as claimed in claim 1, characterized in that: The determination of the customer risk interval according to the customer behavior risk score refers to obtaining the customer behavior risk score R through the customer behavior risk assessment formula; According to the risk preference and business requirements of the financial institution, map the calculated customer behavior risk score R to the set value range [0, 1] through normalization; Set the judgment thresholds R1 and R2, and compare the obtained customer behavior risk score R with R1 and R2 to judge the risk type of the customer: If R < R2, it indicates that this type of customer is a low-risk customer; If R2 ≤ R < R1, it indicates that this type of customer is a medium-risk customer; If R ≥ R1, it indicates that this type of customer is a high-risk customer.

5. The financial product matching processing method based on customer big data and vehicle information as claimed in claim 1, characterized in that: The recommendation of the bank's vehicle loan products to the customer through the visual interface refers to constructing a database containing all the vehicle loan products that can be recommended by the bank. Each vehicle loan product corresponds to a different risk interval. If it is a low-risk customer, recommend the low-interest vehicle loan and fixed-rate loan of large state-owned banks; If it is a medium-risk customer, recommend the medium-interest vehicle loan and floating-rate vehicle loan of joint-stock commercial banks; If it is a high-risk customer, recommend the credit repair vehicle loan of local banks; Select the user-based collaborative filtering algorithm, find the customers with similar behaviors to the current customer according to the similarity formula, and recommend the vehicle loan products preferred by the similar customers to the similar customers through the visual interface.

6. The financial product matching processing method based on customer big data and vehicle information as claimed in claim 1, characterized in that: The encrypted storage of customer and vehicle data refers to selecting the symmetric encryption algorithm AES for data encryption, using a secure key algorithm to generate a 256-bit encryption key, using a secure key management service to store and manage the encryption key, and storing the encrypted data in a distributed file system.

7. A computer device comprising: A memory and a processor; the memory stores a computer program, characterized in that: when the processor executes the computer program, the steps of the financial product matching processing method based on customer big data and vehicle information described in any one of claims 1 to 6 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the financial product matching processing method based on customer big data and vehicle information described in any one of claims 1 to 6 are implemented.

Images