Dynamic performance behavior-based customer and merchant credit evaluation method and system, and medium

Abstract

The invention relates to the technical field of customer and merchant credit evaluation, in particular to a dynamic performance behavior-based customer and merchant credit evaluation method and system and a medium. The method comprises the steps of obtaining historical transaction data of a customer and a merchant; identifying dynamic performance behavior information from historical transaction data of the customer and the merchant by adopting a data mining algorithm; based on the identified dynamic performance behavior information, constructing a dynamic index system of customer and merchant credit evaluation; weighting each index in the dynamic index system to obtain the weight of each index; obtaining latest dynamic performance behavior information of the customer and the merchant; the method comprises the following steps of: extracting required indexes from latest dynamic performance behavior information of a customer service provider based on a dynamic index system of customer service provider credit evaluation, and performing standardization processing on the indexes to obtain standardized values of the indexes; and carrying out weighted calculation on the standardized value of each index and the weight of the corresponding index to obtain the credit score of the customer service provider, so that the accuracy and the real-time performance of the customer service provider credit evaluation can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of merchant credit evaluation, and in particular to a merchant credit evaluation method, system and medium based on dynamic contract performance behavior. Background Art

[0002] Merchants usually refer to businessmen or business representatives who participate in various commercial transactions and economic and trade activities, including but not limited to wholesalers, retailers, distributors, agents, importers and exporters, etc. In today's rapidly developing globalization and international trade, merchants, as the main body of commercial transactions, their credit status is directly related to the safety and efficiency of transactions. Therefore, scientific and accurate credit evaluation of merchants not only helps to reduce transaction risks and improve the efficiency of capital use, but also has important significance for promoting the smooth progress of international trade. For this reason, how to achieve scientific and accurate credit evaluation of merchants has become a hot research topic in investment.

[0003] At present, the existing merchant credit evaluation methods, such as Chinese patent CN114611972A, an artificial intelligence-based merchant credit rating system and method, not only use the merchant's own data as the evaluation basis to evaluate its internal credit risk, but also take into account some critical external factors, such as industry trends, shareholder credit, and news and public opinion, which can effectively evaluate the merchant's external credit risk, and thus can comprehensively evaluate the merchant's credit situation and avoid making wrong investment, financing, cooperation and other business decisions. However, the existing merchant credit evaluation methods still have major technical problems, specifically:

[0004] 1. Lack of comprehensiveness: Existing merchant credit evaluation methods mainly focus on the merchant's basic conditions, financial indicators, innovation and research and development, public credit and honors, as well as analysis of external factors such as industry trends, shareholder credit and news and public opinion, while ignoring dynamic data such as the merchant's performance in the actual business process. As a result, the evaluation results cannot fully and accurately reflect the merchant's credit status, which has an adverse impact on the company's risk control in business decision-making processes such as investment, financing, and cooperation.

[0005] 2. Ignoring real-time nature: Existing merchant credit evaluation methods often focus on the analysis of historical data, but pay insufficient attention to dynamic information such as the merchant's latest performance behavior, market performance, etc., which leads to a certain deviation between the credit evaluation results and the actual situation, making the credit evaluation results lack persuasiveness in some aspects. Summary of the invention

[0006] In response to the above technical problems, the present invention proposes a merchant credit evaluation method, system and medium based on dynamic performance behavior, aiming to effectively explore the credit risk factors in the dynamic performance behavior of merchants, reflect the credit performance of merchants in daily operations and the real-time changes in the credit status of merchants, thereby improving the accuracy and real-time nature of merchant credit evaluation.

[0007] In the first aspect, the present application provides a merchant credit evaluation method based on dynamic performance behavior, comprising the following steps:

[0008] Step 102, obtaining historical transaction data of merchants;

[0009] Step 104, using a data mining algorithm to identify dynamic performance behavior information from the merchant's historical transaction data;

[0010] Step 106, constructing a dynamic indicator system for merchant credit evaluation based on the identified dynamic performance behavior information;

[0011] Step 108, weighting each indicator in the dynamic indicator system to obtain the weight of each indicator;

[0012] Step 110, obtaining the latest dynamic performance information of the merchant;

[0013] Step 112, based on the dynamic indicator system of merchant credit evaluation, extract the required indicators from the latest dynamic performance behavior information of the merchant, and standardize the indicators to obtain standardized values ​​of various indicators;

[0014] Step 114, performing weighted calculation on the standardized value of each indicator and the weight of the corresponding indicator to obtain the credit score of the merchant.

[0015] In some embodiments, in step 104, a data mining algorithm is used to identify dynamic performance behavior information from historical transaction data of merchants, including:

[0016] Step 202, using an association rule mining algorithm to perform association rule mining on the merchant's historical transaction data;

[0017] Step 204, setting a support threshold and a confidence threshold, and screening out frequent item sets and strong association rules based on the set support threshold and confidence threshold;

[0018] Step 206, determining representative dynamic performance behaviors based on frequent item sets and strong association rules;

[0019] Step 208, identifying dynamic fulfillment behavior information from the merchant's historical transaction data based on the determined dynamic fulfillment behavior.

[0020] In some embodiments, in step 202, association rule mining is performed on the historical transaction data of merchants using an association rule mining algorithm, including:

[0021] Based on the historical transaction data of merchants, the merchants are clustered using clustering algorithms;

[0022] Analyze the feature distribution in different clusters and identify groups of merchants with similar transaction patterns;

[0023] The association rule mining algorithm is used to mine the historical transaction data of various merchants.

[0024] In some embodiments, in step 104, using a data mining algorithm to identify dynamic performance behavior information from historical transaction data of merchants also includes:

[0025] Periodically obtain the latest transaction data of customers;

[0026] Based on the latest transaction data of the merchant, steps 202 to 206 are executed to update representative dynamic performance behaviors.

[0027] In some embodiments, in step 108, each indicator in the dynamic indicator system is weighted to obtain the weight of each indicator, including:

[0028] Step 302, using the principal component analysis method to preliminarily assign weights to each indicator in the dynamic indicator system to obtain the first weight of each indicator;

[0029] Step 304, using the expert scoring method to perform secondary weighting on each indicator in the dynamic indicator system to obtain a second weight of each indicator;

[0030] Step 306: The first weight and the second weight are fused, and the obtained fusion weight is the final weight of each indicator.

[0031] In some embodiments, in step 302, the principal component analysis method is used to preliminarily assign weights to each indicator in the dynamic indicator system to obtain a first weight of each indicator, including:

[0032] Calculate the covariance matrix between the indicators in the dynamic indicator system;

[0033] Solve for the eigenvalues ​​and eigenvectors of the covariance matrix;

[0034] According to the size of the eigenvalue, the eigenvalues ​​are arranged in descending order, and the eigenvectors corresponding to the first several eigenvalues ​​are selected as the principal components;

[0035] Construct a principal component matrix based on the selected principal components, and calculate the principal component score of each indicator according to the principal component matrix;

[0036] Based on the principal component score of each indicator, the first weight of each indicator is calculated.

[0037] In some embodiments, in step 304, the expert scoring method is used to perform secondary weighting on each indicator in the dynamic indicator system to obtain a second weight of each indicator, including:

[0038] Experts will compare each indicator pairwise and evaluate the relative importance of each indicator based on a 1-9 scale;

[0039] Construct a judgment matrix for each indicator based on its relative importance;

[0040] Calculate the maximum eigenvalue and corresponding eigenvector of each judgment matrix, the eigenvector corresponding to the maximum eigenvalue contains the relative weights of various indicators;

[0041] The eigenvector corresponding to the maximum eigenvalue is normalized to obtain the second weight of each indicator.

[0042] In some embodiments, in step 306, the first weight and the second weight are fused, and the fused weight obtained is the final weight of each indicator, including:

[0043] Taking minimizing the weighted average of the first weight and the second weight as the objective function;

[0044] Based on the objective function, a linear programming model is used to find the optimal weight combination of various indicators;

[0045] According to the optimal weight combination of various indicators, the fusion weight of each indicator is calculated as the final weight of each indicator.

[0046] Secondly, this application provides a merchant credit evaluation system based on dynamic performance behavior, including:

[0047] The data acquisition module is used to obtain the historical transaction data of merchants and the latest dynamic performance behavior information of merchants;

[0048] An information identification module is used to identify dynamic performance behavior information from the merchant's historical transaction data using a data mining algorithm;

[0049] The indicator system construction module is used to build a dynamic indicator system for merchant credit evaluation based on the identified dynamic performance behavior information;

[0050] The weighting module is used to weight each indicator in the dynamic indicator system and obtain the weight of each indicator;

[0051] The standardization processing module is used to extract the required indicators from the latest dynamic performance behavior information of the merchants based on the dynamic indicator system of the merchant credit evaluation, and standardize the indicators to obtain the standardized values ​​of each indicator;

[0052] The credit score output module is used to perform weighted calculation on the standardized value of each indicator and the weight of the corresponding indicator to obtain the credit score of the merchant.

[0053] In a third aspect, a computer-readable storage medium stores a computer program, which, when executed by a processor, implements the aforementioned customer credit evaluation method based on dynamic performance behavior.

[0054] The beneficial technical effects of the present invention include at least: adopting a credit evaluation method, system and medium for merchants based on dynamic performance behavior, firstly, by clustering and mining association rules of transaction data of different merchant groups, the behavior patterns of different merchant groups can be identified to meet the personalized needs of different enterprises for business decisions such as investment, financing, and cooperation; secondly, by using an association rule mining algorithm to mine the historical transaction data of merchants, frequent item sets and strong association rules are screened out, and based on the frequent item sets and strong association rules, representative dynamic performance behaviors are determined, which can more comprehensively and accurately identify the dynamic performance behavior information that affects the credit of merchants, making subsequent credit evaluation more accurate, and at the same time, based on the latest transaction data of merchants, representative dynamic performance behaviors are updated, which can realize real-time monitoring of the credit status of merchants, and use this as the basis for subsequent credit evaluation of merchants, ensuring that the determined dynamic performance behaviors can more accurately reflect the dynamic development trend of the market and industry; then, the first weight of each indicator is obtained by using the principal component analysis method for preliminary weighting, and the second weighting of each indicator is obtained by using the expert scoring method for secondary weighting. The first weight and the second weight are then integrated, and the integrated weights are used as the final weights of each indicator. This can comprehensively consider multiple factors, including not only the correlation between the indicators in the dynamic indicator system, but also the judgment of experts, and provide a scientific, reasonable and practical solution for the weight allocation of each indicator in the dynamic indicator system, which is helpful to further improve the accuracy of the credit score of subsequent merchants. In the process of weighting by principal component analysis, by minimizing the weighted average of the first weight and the second weight as the objective function, the focus can be placed on reducing risks and adverse effects, and those merchants with poor performance in key indicators but good overall credit can be more effectively judged; finally, based on the latest dynamic performance behavior information of the merchant, the standardized values ​​of each indicator and the weights of the corresponding indicators are weighted to obtain the credit score of the merchant, effectively digging out the credit risk factors in the dynamic performance behavior of the merchant, reflecting the credit performance of the merchant in daily operations and the real-time changes of the merchant's credit status, thereby improving the accuracy and real-time nature of the merchant's credit evaluation, and improving the quality and efficiency of the credit evaluation, which has important practical value for enterprises and investors.

[0055] Other features and advantages of the present invention will be disclosed in detail in the following specific embodiments and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The present invention will be further described below in conjunction with the accompanying drawings:

[0057] Figure 1 The present invention is a flowchart of a merchant credit evaluation method based on dynamic performance behavior according to an embodiment of the present invention.

[0058] Figure 2The figure is a schematic diagram of the structure of a merchant credit evaluation system based on dynamic performance behavior according to an embodiment of the present invention. DETAILED DESCRIPTION

[0059] The technical solutions of the embodiments of the present invention are explained and described below in conjunction with the drawings of the embodiments of the present invention, but the following embodiments are only preferred embodiments of the present invention, not all. Based on the embodiments in the implementation mode, other embodiments obtained by those skilled in the art without creative work are all within the protection scope of the present invention.

[0060] In the following description, terms such as "inside", "outside", "up", "down", "left", "right", etc. that indicate directions or positional relationships are only used to facilitate the description of the embodiments and simplify the description, and do not indicate or imply that the referred device or element must have a specific direction, be constructed and operate in a specific direction. Therefore, it should not be understood as a limitation of the present invention.

[0061] Please see attached Figure 1 , Figure 1 A flow chart of a merchant credit evaluation method based on dynamic performance behavior provided in one embodiment of this specification is shown.

[0062] like Figure 1 As shown, the merchant credit evaluation method based on dynamic performance behavior may at least include the following steps:

[0063] Step 102, obtaining the merchant's historical transaction data.

[0064] Furthermore, the historical transaction data of merchants can be cleaned, missing values ​​can be processed, errors can be corrected, and formats can be standardized to ensure the quality of the data.

[0065] Step 104, using a data mining algorithm to identify dynamic fulfillment behavior information from the merchant's historical transaction data.

[0066] Specifically, in this embodiment, in step 104, a data mining algorithm is used to identify dynamic performance behavior information from the merchant's historical transaction data, including:

[0067] Step 202: Use an association rule mining algorithm to perform association rule mining on the historical transaction data of merchants.

[0068] The association rule mining algorithm includes but is not limited to the Apriori algorithm, the FP-growth algorithm, etc. The association rule mining algorithm aims to discover the relationship between items in the data set. For example, in the historical transaction data set of merchants, each transaction behavior can be an item.

[0069] Furthermore, in this embodiment, in step 202, using an association rule mining algorithm to perform association rule mining on the historical transaction data of merchants may also include:

[0070] Based on the historical transaction data of merchants, the merchants are clustered using clustering algorithms;

[0071] Analyze the feature distribution in different clusters and identify groups of merchants with similar transaction patterns;

[0072] The association rule mining algorithm is used to mine the historical transaction data of various merchants.

[0073] Among them, based on the historical transaction data of merchants, the implementation method of clustering merchants using clustering algorithm is as follows:

[0074] First, select a clustering algorithm: In this embodiment, a suitable clustering algorithm can be selected according to the characteristics of the merchant's historical transaction data and the clustering target, and this embodiment does not limit this. For example:

[0075] If the transaction patterns of merchants in the transaction data are relatively consistent, that is, the differences in transaction behaviors between merchants are mainly reflected in transaction frequency and amount, and we want to identify groups of merchants with similar transaction behaviors, then the K-means clustering algorithm may be a better choice. The K-means algorithm is simple and efficient, and can effectively find clusters of similar sizes;

[0076] If the transaction behaviors of merchants in the transaction data vary greatly, some of them may be based on specific events or seasonal fluctuations. Such historical transaction data may form clusters of arbitrary shapes. In this case, the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering algorithm is more appropriate. The DBSCAN algorithm does not require the pre-specified number of clusters, can identify clusters of arbitrary shapes, and can handle noise points. The DBSCAN algorithm determines the formation of clusters by calculating the density between data points, so it is particularly suitable for discovering merchant groups in different density areas.

[0077] Secondly, determine the clustering parameters: For the K-means algorithm, the number of clusters can be determined based on features such as transaction frequency and amount. The K-means algorithm will try to group customers with similar transaction patterns into the same cluster; for the DBSCAN algorithm, it is necessary to determine the neighborhood radius and the minimum number of points, and then determine the formation of clusters by calculating the density between data points.

[0078] Finally, execute the clustering algorithm: use the selected clustering algorithm to cluster merchants based on their historical transaction data.

[0079] Among them, the method of analyzing the feature distribution in different clusters and identifying merchant groups with similar transaction patterns is as follows:

[0080] First, observe the distribution of transaction behavior characteristics of each cluster, and use internal indicators (such as silhouette coefficient) to evaluate the cohesion and separation of each cluster to determine the quality of the cluster;

[0081] Secondly, the similarities within and between clusters are compared by internal and external distances to identify the transaction patterns of different clusters;

[0082] Finally, the clustering results are combined with business background knowledge to interpret the characteristics of the clustering results. For example, a cluster with generally high transaction amounts may indicate a high-value customer group, and a cluster with a significantly lower transaction frequency than other groups or a significant decrease in transaction amounts may indicate potential lost customers.

[0083] By clustering and mining association rules on the transaction data of different merchant groups, we can identify the behavior patterns of different merchant groups to meet the personalized needs of different companies for business decisions such as investment, financing, and cooperation.

[0084] Step 204: set a support threshold and a confidence threshold, and filter out frequent item sets and strong association rules based on the set support threshold and confidence threshold.

[0085] Among them, support is the frequency of an item set appearing in all transactions, and an item set refers to the set of all items in the data set. The support threshold is a user-specified minimum frequency used to filter out unimportant item sets.

[0086] The implementation method of filtering frequent itemsets can be:

[0087] 1. The Apriori algorithm calculates the support of item sets by scanning the data set multiple times, and prunes the item sets that do not meet the support, so as to screen out frequent item sets.

[0088] 2. The FP-growth algorithm uses a data structure called a prefix tree to efficiently find frequent itemsets.

[0089] Among them, confidence is the conditional probability of the occurrence of consequence when the antecedent occurs. The confidence threshold is a user-specified minimum confidence value used to filter out unimportant association rules.

[0090] The implementation method of screening out strong association rules can be:

[0091] First, for each frequent item set, all possible association rules are generated;

[0092] Then, the confidence of each association rule is calculated and compared with the confidence threshold, and those association rules whose confidence is not less than the confidence threshold are retained to screen out strong association rules.

[0093] Step 206, based on the frequent item sets and strong association rules, determine representative dynamic fulfillment behaviors.

[0094] For example, frequent item sets and strong association rules may be that certain groups of merchants often purchase specific goods at the same time, or certain groups of merchants often conduct transactions at a specific time, etc.

[0095] For example, the representative dynamic performance behaviors determined in this embodiment include, but are not limited to, bidding winning behaviors, contract signing and performance behaviors, payment and collection behaviors, complaints and claims behaviors, etc.

[0096] Furthermore, in this embodiment, in step 104, using a data mining algorithm to identify dynamic performance behavior information from the merchant's historical transaction data also includes:

[0097] Periodically obtain the latest transaction data of customers;

[0098] Based on the latest transaction data of the merchants, steps 202 to 206 are executed to update representative dynamic performance behaviors, thereby ensuring that the determined dynamic performance behaviors can more accurately reflect the dynamic development trends of the market and the industry.

[0099] Step 208, identifying dynamic fulfillment behavior information from the merchant's historical transaction data based on the determined dynamic fulfillment behavior.

[0100] For example, the dynamic performance behavior information identified in this embodiment includes, but is not limited to, winning bid section ID information, contract number, customer contract signing information, financial payment information, financial collection information, complaint and claim information, etc.

[0101] This embodiment uses an association rule mining algorithm to perform association rule mining on the historical transaction data of merchants, screens out frequent item sets and strong association rules, and determines representative dynamic performance behaviors based on the frequent item sets and strong association rules. It can more comprehensively and accurately identify the dynamic performance behavior information that affects the credit of merchants, making subsequent credit evaluation more accurate. At the same time, based on the latest transaction data of merchants, representative dynamic performance behaviors are updated, which can achieve real-time monitoring of the credit status of merchants. This serves as the basis for subsequent credit evaluation of merchants, ensuring that the determined dynamic performance behaviors can more accurately reflect the dynamic development trends of the market and industry. Moreover, this method emphasizes data-based, extracts valuable laws through data mining technology, reduces subjective judgment, and greatly improves the scientific nature of decision-making.

[0102] Step 106, constructing a dynamic indicator system for merchant credit evaluation based on the identified dynamic performance behavior information.

[0103] By way of example, the dynamic indicator system for merchant credit evaluation constructed in this embodiment includes but is not limited to the number of successful bids, expert scores, years of cooperation, successful bid rate, time from successful bid to contract signing, quality of contract performance, time from contract signing to completion of contract performance, time from contract performance to completion of payment, cooperation stability, customer satisfaction and other indicators.

[0104] Step 108, weighting each indicator in the dynamic indicator system to obtain the weight of each indicator.

[0105] Specifically, in this embodiment, in step 108, each indicator in the dynamic indicator system is weighted to obtain the weight of each indicator, including:

[0106] Step 302: Use the principal component analysis method to preliminarily assign weights to each indicator in the dynamic indicator system to obtain the first weight of each indicator.

[0107] Furthermore, in this embodiment, in step 302, the principal component analysis method is used to preliminarily assign weights to each indicator in the dynamic indicator system to obtain the first weight of each indicator, including:

[0108] Step 402, calculating the covariance matrix between various indicators in the dynamic indicator system.

[0109] It can be understood that the covariance matrix describes the correlation between various indicators. For a vector X containing n indicators, n ], the calculation formula of the covariance matrix Cov(X) can be:

[0110]

[0111] Among them, n represents the total number of indicators, x i represents the value of the i-th index, μ x Indicates the mean of the currently calculated indicator X.

[0112] Step 404, solving the eigenvalues ​​and eigenvectors of the covariance matrix.

[0113] Specifically, the eigenvalues ​​and eigenvectors are obtained by solving the characteristic equation of the covariance matrix. The characteristic equation can be expressed as:

[0114] det(Cov(X)-λI)=0

[0115] Among them, λ represents the eigenvalue and I represents the identity matrix.

[0116] Step 406: Arrange the eigenvalues ​​in descending order according to their magnitudes, and select the eigenvectors corresponding to the first several eigenvalues ​​as principal components.

[0117] The selection of the number of the first several eigenvalues ​​depends on the variance ratio of the data we hope to explain. In credit evaluation, we can pay more attention to factors that have a direct impact on credit changes, or we can use the "80 / 20 rule" to select features that can explain at least 80% of the data variance. This embodiment does not limit this.

[0118] Step 408, constructing a principal component matrix based on the selected principal components, and calculating the principal component score of each indicator according to the principal component matrix.

[0119] Among them, the principal component score is the projection of the original dynamic indicator system in the new dynamic indicator system, that is, the position of the original indicator data in the new coordinate system. These coordinates represent the main change direction of the original indicator data. For each indicator, its principal component score y can be expressed as:

[0120] y=Z·W

[0121] Among them, Z represents the original indicator data matrix, and W represents the principal component matrix composed of the eigenvectors corresponding to the selected principal components.

[0122] Step 410, based on the principal component score of each indicator, calculate the first weight of each indicator.

[0123] It can be understood that the coefficient of the principal component can be used as the first weight of each indicator. i It can be obtained by normalizing the components of the eigenvector and then taking the average:

[0124]

[0125] Among them, k represents the eigenvector v in the principal component matrix j The total number, v ji Denotes the feature vector v j The i-th component of .

[0126] Step 304, using the expert scoring method to perform secondary weighting on each indicator in the dynamic indicator system to obtain a second weight of each indicator.

[0127] Furthermore, in this embodiment, in step 304, the expert scoring method is used to perform secondary weighting on each indicator in the dynamic indicator system to obtain a second weight of each indicator, including:

[0128] In step 502, experts compare each indicator in pairs and evaluate the relative importance of each indicator according to a 1-9 scale.

[0129] Usually, on a 1-9 scale, 1 means that two indicators are equally important, and 9 means that one indicator is much more important than the other.

[0130] Step 504: construct a judgment matrix for each indicator based on the relative importance of each indicator.

[0131] Among them, for the pairwise comparison of n indicators, an n×n judgment matrix will be obtained, the diagonal of the judgment matrix is ​​1, and the off-diagonal elements a ij Indicates the importance of indicator i relative to indicator j.

[0132] Step 506, calculating the maximum eigenvalue and the corresponding eigenvector of each judgment matrix, wherein the eigenvector corresponding to the maximum eigenvalue includes the relative weight of each indicator.

[0133] Mathematical software (such as MATLAB, Python, etc.) or eigenvalue calculation tools can be used to calculate the maximum eigenvalue of the judgment matrix and its corresponding eigenvector.

[0134] Step 508: normalize the eigenvector corresponding to the maximum eigenvalue to obtain the second weight of each indicator.

[0135] It can be understood that the eigenvector corresponding to the maximum eigenvalue is normalized, that is, the sum of the elements in the eigenvector is set to 1, and the normalized eigenvector is the second weight of each indicator.

[0136] Step 306: The first weight and the second weight are fused, and the obtained fusion weight is the final weight of each indicator.

[0137] Furthermore, in this embodiment, in step 306, the first weight and the second weight are fused, and the obtained fusion weight is the final weight of each indicator, including:

[0138] Step 602: minimizing the weighted average of the first weight and the second weight is used as the objective function.

[0139] The objective function can be expressed as:

[0140]

[0141] Among them, x i1 and x i2 are decision variables, namely, the first and second weights representing the optimal i-th indicator, w i1 represents the first weight of the i-th indicator obtained in the previous step, w i2 It represents the second weight of the i-th indicator obtained by rebalancing in the previous step.

[0142] It is understandable that an important purpose of merchant credit evaluation is risk management. If a merchant performs poorly in certain key indicators, then even if the merchant's overall weighted score is high, it should be considered a high-risk merchant. Therefore, this embodiment can focus on reducing risks and adverse effects by minimizing the weighted average of the first weight and the second weight as the objective function, and can more effectively determine those merchants who perform poorly in key indicators but have good overall credit.

[0143] Step 604: Based on the objective function, a linear programming model is used to find the optimal weight combination of various indicators.

[0144] Specifically, first determine the constraints, which can be expressed as:

[0145]

[0146] x i1 ,x i2 ≥0

[0147] w i1 ,w i2 ≥0

[0148] After having the objective function and constraints, we can then use a linear programming solver to solve the optimal weight combination. Commonly used linear programming solvers include MATLAB's linprog function, Python's scipy.optimize.linprog function, Excel's solver, etc. Input the objective function and constraints into the selected linear programming solver, and the solver will output the optimal decision variable x i1 and x i2 .

[0149] Step 606, according to the optimal weight combination of each indicator, the fusion weight of each indicator is obtained as the final weight of each indicator.

[0150] For example, the optimal decision variable x corresponding to each indicator i1 and x i2 By adding and taking the average, we can get the fusion weight of each indicator, which is used as the final weight of each indicator.

[0151] This embodiment uses the principal component analysis method to perform preliminary weighting to obtain the first weight of each indicator, uses the expert scoring method to perform secondary weighting to obtain the second weight of each indicator, and then fuses the first weight and the second weight to obtain the fused weight as the final weight of each indicator. It can comprehensively consider multiple factors, including not only the correlation between the indicators in the dynamic indicator system, but also the judgment of experts, and provide a scientific, reasonable and practical solution for the weight allocation of various indicators in the dynamic indicator system, which is helpful to further improve the accuracy of credit scores of subsequent customers.

[0152] Step 110, obtaining the latest dynamic performance information of the merchant;

[0153] Step 112, based on the dynamic indicator system of merchant credit evaluation, extract the required indicators from the merchant's latest dynamic performance behavior information, and standardize the indicators to obtain standardized values ​​of various indicators.

[0154] It is understandable that, since different indicators may have different dimensions and numerical ranges, the indicators need to be standardized to ensure that their contributions to the score are equivalent. Commonly used standardization methods include minimum-maximum standardization and Z-score standardization, which are not limited in this embodiment.

[0155] Step 114, performing weighted calculation on the standardized value of each indicator and the weight of the corresponding indicator to obtain the credit score of the merchant.

[0156] See also Figure 2 , Figure 2 A schematic diagram of the structure of a merchant credit evaluation system based on dynamic performance behavior provided for one embodiment of this specification.

[0157] As shown in the figure, the merchant credit evaluation system based on dynamic performance behavior may at least include a data acquisition module 1 and an information identification module 2, wherein:

[0158] Data acquisition module 1, used to obtain historical transaction data of merchants and the latest dynamic performance behavior information of merchants;

[0159] Information identification module 2, used to identify dynamic performance behavior information from historical transaction data of merchants using data mining algorithms;

[0160] Index system construction module 3 is used to construct a dynamic index system for merchant credit evaluation based on the identified dynamic performance behavior information;

[0161] The weighting module 4 is used to weight each indicator in the dynamic indicator system to obtain the weight of each indicator;

[0162] The standardization processing module 5 is used to extract the required indicators from the latest dynamic performance behavior information of the merchants based on the dynamic indicator system of the merchant credit evaluation, and to standardize the indicators to obtain the standardized values ​​of each indicator;

[0163] The credit score output module 6 is used to perform weighted calculation on the standardized value of each indicator and the weight of the corresponding indicator to obtain the credit score of the merchant.

[0164] It can be understood that the technical concept of the merchant credit evaluation system based on dynamic performance behavior provided in this embodiment is similar to the technical concept of the aforementioned merchant credit evaluation method based on dynamic performance behavior, and this embodiment will not be repeated here.

[0165] Another embodiment of the present specification provides a computer-readable storage medium, which stores instructions, and when the instructions are executed on a computer or a processor, the computer or the processor executes one or more steps in the above-mentioned embodiment of the method for evaluating the credit of a merchant based on dynamic performance behavior. If the components of the above-mentioned electronic device are implemented in the form of software functional units and used as independent downstream task predictions or uses, they can be stored in a computer-readable storage medium.

[0166] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loading and executing computer program instructions on a computer, the process or function according to the embodiment of this specification is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted by a computer-readable storage medium. The computer instructions can be transmitted from a website site, a computer, a server or a data center to another website site, a computer, a server or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (Digital Subscriber Line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server, a data center, etc. that contains one or more available media integrations. Available media may be magnetic media (eg, floppy disks, hard disks, magnetic tapes), optical media (eg, digital versatile discs (DVD)), or semiconductor media (eg, solid state disks (SSD)).

[0167] The above descriptions are only preferred embodiments disclosed in this application and descriptions of the technical principles used. Those skilled in the art should understand that the scope of protection involved in this disclosure is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in this disclosure (but not limited to) to form a technical solution.

[0168] In addition, although each operation is described in a specific order, this should not be understood as requiring these operations to be performed in the specific order shown or in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be interpreted as limiting the scope of the present disclosure. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

Claims

1. A merchant credit evaluation method based on dynamic performance behavior, characterized in that: The following steps are involved: Step 102, obtaining historical transaction data of merchants; Step 104, using a data mining algorithm to identify dynamic performance behavior information from the merchant's historical transaction data; Step 106, constructing a dynamic indicator system for merchant credit evaluation based on the identified dynamic performance behavior information; Step 108, weighting each indicator in the dynamic indicator system to obtain the weight of each indicator; Step 110, obtaining the latest dynamic performance information of the merchant; Step 112, based on the dynamic indicator system of merchant credit evaluation, extract the required indicators from the latest dynamic performance behavior information of the merchant, and standardize the indicators to obtain standardized values ​​of various indicators; Step 114, performing weighted calculation on the standardized value of each indicator and the weight of the corresponding indicator to obtain the credit score of the merchant.

2. The merchant credit evaluation method based on dynamic performance behavior as claimed in claim 1, characterized in that: In step 104, a data mining algorithm is used to identify dynamic performance behavior information from the merchant's historical transaction data, including: Step 202, using an association rule mining algorithm to perform association rule mining on the merchant's historical transaction data; Step 204, setting a support threshold and a confidence threshold, and screening out frequent item sets and strong association rules based on the set support threshold and confidence threshold; Step 206, determining representative dynamic performance behaviors based on frequent item sets and strong association rules; Step 208, identifying dynamic fulfillment behavior information from the merchant's historical transaction data based on the determined dynamic fulfillment behavior.

3. The merchant credit evaluation method based on dynamic performance behavior as claimed in claim 2, characterized in that: In step 202, association rule mining is performed on the historical transaction data of merchants using an association rule mining algorithm, including: Based on the historical transaction data of merchants, the merchants are clustered using clustering algorithms; Analyze the feature distribution in different clusters and identify groups of merchants with similar transaction patterns; The association rule mining algorithm is used to mine the historical transaction data of various merchants.

4. The merchant credit evaluation method based on dynamic performance behavior as claimed in claim 2, characterized in that: In step 104, a data mining algorithm is used to identify dynamic performance behavior information from the merchant's historical transaction data, which also includes: Periodically obtain the latest transaction data of customers; Based on the latest transaction data of the merchant, steps 202 to 206 are executed to update representative dynamic performance behaviors.

5. The merchant credit evaluation method based on dynamic performance behavior as claimed in claim 1, characterized in that: In step 108, each indicator in the dynamic indicator system is weighted to obtain the weight of each indicator, including: Step 302, using the principal component analysis method to preliminarily assign weights to each indicator in the dynamic indicator system to obtain the first weight of each indicator; Step 304, using the expert scoring method to perform secondary weighting on each indicator in the dynamic indicator system to obtain a second weight of each indicator; Step 306: The first weight and the second weight are fused, and the obtained fusion weight is the final weight of each indicator.

6. The merchant credit evaluation method based on dynamic performance behavior as claimed in claim 5, characterized in that: In step 302, the principal component analysis method is used to preliminarily assign weights to the various indicators in the dynamic indicator system to obtain the first weights of the various indicators, including: Calculate the covariance matrix between the indicators in the dynamic indicator system; Solve for the eigenvalues ​​and eigenvectors of the covariance matrix; According to the size of the eigenvalue, the eigenvalues ​​are arranged in descending order, and the eigenvectors corresponding to the first several eigenvalues ​​are selected as the principal components; Construct a principal component matrix based on the selected principal components, and calculate the principal component score of each indicator according to the principal component matrix; Based on the principal component score of each indicator, the first weight of each indicator is calculated.

7. The merchant credit evaluation method based on dynamic performance behavior as claimed in claim 5, characterized in that: In step 304, the expert scoring method is used to perform secondary weighting on each indicator in the dynamic indicator system to obtain the second weight of each indicator, including: Experts will compare each indicator pairwise and evaluate the relative importance of each indicator based on a 1-9 scale; Construct a judgment matrix for each indicator based on its relative importance; Calculate the maximum eigenvalue and corresponding eigenvector of each judgment matrix, the eigenvector corresponding to the maximum eigenvalue contains the relative weights of various indicators; The eigenvector corresponding to the maximum eigenvalue is normalized to obtain the second weight of each indicator.

8. The merchant credit evaluation method based on dynamic performance behavior as claimed in claim 5, characterized in that: In step 306, the first weight and the second weight are fused, and the obtained fusion weight is the final weight of each indicator, including: Taking minimizing the weighted average of the first weight and the second weight as the objective function; Based on the objective function, a linear programming model is used to find the optimal weight combination of various indicators; According to the optimal weight combination of various indicators, the fusion weight of each indicator is calculated as the final weight of each indicator.

9. The merchant credit evaluation system based on dynamic performance behavior is characterized by: include: The data acquisition module is used to obtain the historical transaction data of merchants and the latest dynamic performance behavior information of merchants; An information identification module is used to identify dynamic performance behavior information from the merchant's historical transaction data using a data mining algorithm; The indicator system construction module is used to build a dynamic indicator system for merchant credit evaluation based on the identified dynamic performance behavior information; The weighting module is used to weight each indicator in the dynamic indicator system and obtain the weight of each indicator; The standardization processing module is used to extract the required indicators from the latest dynamic performance behavior information of the merchants based on the dynamic indicator system of the merchant credit evaluation, and standardize the indicators to obtain the standardized values ​​of each indicator; The credit score output module is used to perform weighted calculation on the standardized value of each indicator and the weight of the corresponding indicator to obtain the credit score of the merchant.

10. A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the customer credit evaluation method based on dynamic performance behavior as described in any one of claims 1 to 8.

Images