Supply chain financial transaction security early warning method and system

Abstract

The invention relates to the technical field of data processing, and relates to a supply chain financial transaction security early warning method and system, and the method comprises the steps: obtaining financial data of a supply chain and a corresponding risk score, dividing the financial data into continuous data and discrete data, and calculating the importance coefficient of the continuous data and the discrete data; the importance coefficient is positively correlated with a Spearman level correlation coefficient of the risk score, and is inversely correlated with a Pearson correlation coefficient in the continuous data, standardized mutual information in the discrete data and a point double-column correlation coefficient between the continuous data and the discrete data. And performing risk type probability prediction on the financial data by using an improved XGBoost model, wherein the risk type corresponding to the maximum value of the probability is the final risk type of the financial data. According to the method, the XGBoost model is improved, so that the problem of inaccurate supply chain financial transaction risk assessment is solved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology. More specifically, the present invention relates to a supply chain financial transaction security early warning method and system. Background Art

[0002] With the continuous development of supply chain finance, transactions between enterprises are becoming more frequent, capital flows are becoming more complicated, and the risks in financial transactions are also increasing. Especially in supply chain finance, the transaction chain between different enterprises is long and involves many links, from raw material supply to production and manufacturing to sales terminals. The break of the capital chain in any link will affect the stability of the entire supply chain. Therefore, real-time monitoring of risk factors in the transaction process and effective early warning have become the key to ensuring the security of supply chain finance. This can not only improve the overall efficiency of the supply chain, but also reduce the economic losses caused by risks to financial institutions and enterprises, which has important practical significance and application prospects.

[0003] At present, the security warning of supply chain financial transactions mainly relies on the analysis of corporate financial data and credit rating assessment. However, traditional analysis methods are mostly based on static financial statements and historical data, which are difficult to reflect the dynamic changes of enterprises in real time, especially when the external market environment changes, potential risks may be ignored. In addition, the financial data types of different enterprises are different, some data are continuous, and some data are discrete. How to effectively process these data and integrate various types of information has become a technical problem.

[0004] Faced with such a complex data structure, traditional risk assessment models find it difficult to accurately capture the risk signals hidden behind the data. Among them, XGBoost is an efficient gradient boosting decision tree algorithm model that iteratively trains multiple decision trees to gradually reduce the prediction error of the previous round of models. Each round of iteration adjusts the model according to the prediction error of the previous round, so that the new decision tree pays more attention to difficult-to-predict samples. However, XGBoost is sensitive to hyperparameters. Different hyperparameter configurations may lead to significant differences in model effects. It often requires a complex tuning process to ensure that the model can achieve optimal performance on a specific data set. Especially when applied to different data sets, a large number of experiments are often required to find the optimal parameter configuration. Therefore, the uncertainty of parameter configuration often leads to inaccurate risk assessment of supply chain financial transactions. Summary of the invention

[0005] In order to solve the problem that the uncertainty of parameter configuration mentioned in the above background technology often leads to inaccurate risk assessment of supply chain financial transactions, the present invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a supply chain financial transaction security early warning method, comprising:

[0007] Obtain financial data of the supply chain and its corresponding risk score, and divide the financial data into continuous data and discrete data; calculate the importance coefficient of the financial data, the importance coefficient is positively correlated with the Spearman rank correlation coefficient of the risk score, and is negatively correlated with the Pearson correlation coefficient in continuous data, the standardized mutual information in discrete data, and the point biserial correlation coefficient between continuous data and discrete data; use the improved XGBoost model to monitor the risk of the financial data to obtain the final probability of each risk type to which the financial data belongs; wherein the improved XGBoost model includes: calculating the weight of each classification decision tree in the XGBoost model, the weight is positively correlated with the importance coefficient, and is negatively correlated with the number of financial data assigned to each classification decision tree; weighted averaging the weight of each classification decision tree with the probability of each risk type output by each classification decision tree to obtain the final probability of each risk type to which the financial data belongs, and the risk type corresponding to the maximum value of the final probability is the final risk type of the financial data.

[0008] The above technical solution calculates the importance coefficient of financial data by utilizing a variety of correlation analysis methods, thereby optimizing and improving the weight distribution of classification decision trees in the XGBoost model. At the same time, it reduces the bias of a single decision tree through weighted averaging, thereby improving the accuracy and robustness of the prediction of various risk types of financial data.

[0009] Furthermore, the importance coefficient C of the i-th financial data is:

[0010]

[0011] In the formula, r i represents the Spearman rank correlation coefficient between the ith financial data and its corresponding risk score, α represents the weight factor, ln represents the logarithmic function with base e, r represents the Pearson correlation coefficient in continuous data, NMI represents the standardized mutual information in discrete data, and R represents the point-biserial correlation coefficient between continuous and discrete data.

[0012] The above technical solution accurately evaluates the importance coefficient of each financial data by introducing a calculation formula based on the Spearman rank correlation coefficient, Pearson correlation coefficient, standardized mutual information and point-biserial correlation coefficient. By using logarithmic function smoothing and combining the weight factor α, a comprehensive balance of continuous and discrete data is achieved, thereby more accurately capturing the complex relationship between financial data and risk scores.

[0013] Furthermore, the weight D of the k-th classification decision tree k for:

[0014]

[0015] In the formula, C i represents the importance coefficient of the i-th financial data, and M represents the number of financial data assigned to the k-th classification decision tree.

[0016] The above technical solution ensures that the weight of each decision tree reflects the importance of the data it processes by associating the weight of the classification decision tree with the importance coefficient of the financial data. This not only strengthens the focus on high-importance data, but also reduces the negative impact of low-quality decision trees on the overall performance of the model.

[0017] Furthermore, the continuous data includes: debt-to-asset ratio, current ratio, quick ratio, order cycle, transaction amount standard deviation and transaction amount; the discrete data includes: number of transactions, number of overdue payments, customer credit score and number of historical defaults.

[0018] Furthermore, the specific process of risk monitoring of the financial data is as follows:

[0019] Construct an XGBoost model, train the XGBoost model to obtain a trained XGBoost model; input the financial data into the trained XGBoost model to obtain a final probability of the risk type to which the financial data belongs.

[0020] Furthermore, the risk types include: a high risk type, a medium risk type, and a low risk type.

[0021] Furthermore, it also includes using spline interpolation method to preprocess the financial data.

[0022] In a second aspect, the present invention provides a supply chain financial transaction security early warning system, comprising a memory and a processor, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a supply chain financial transaction security early warning method described in any one of the above is implemented.

[0023] The beneficial effects of the present invention are:

[0024] The present invention divides financial data into continuous and discrete data, and combines the Spearman rank correlation coefficient, Pearson correlation coefficient, standardized mutual information and point biserial correlation coefficient to determine the importance coefficient of the data, thereby improving the feature selection effect of financial data. At the same time, the improved XGBoost model performs weighted averaging on the probability of each risk type according to the importance coefficient and the weight of the classification decision tree, improving the classification accuracy of the model and the ability to identify different risk types, and effectively enhancing the risk monitoring capability of the supply chain financial system. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0026] Figure 1 is a flow chart schematically illustrating a supply chain financial transaction security early warning method according to an embodiment of the present invention;

[0027] Figure 2 is a structural block diagram schematically illustrating a supply chain financial transaction security early warning system according to an embodiment of the present invention; DETAILED DESCRIPTION

[0028] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0029] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0030] An embodiment of a supply chain financial transaction security early warning method.

[0031] like Figure 1 As shown, a flow chart of a supply chain financial transaction security early warning method according to an embodiment of the present invention includes the following steps:

[0032] S1: Obtain financial data of the supply chain and its corresponding risk score, and divide the financial data into continuous data and discrete data.

[0033] In this embodiment, a set of financial data of one or more enterprises and their corresponding risk scores are obtained from a supply chain finance system through an application programming interface, and the risk scores can be obtained through manual calibration. Of course, the financial data can also be input into a machine learning model and then output; the financial data is divided into continuous data and discrete data based on the continuity of the values; and the financial data is preprocessed using the spline interpolation method; among them, the continuous data includes: debt-to-asset ratio, current ratio, quick ratio, order cycle, transaction amount standard deviation and transaction amount; the discrete data includes: number of transactions, number of overdue payments, customer credit score and number of historical defaults.

[0034] S2: Calculate the importance coefficient of financial data.

[0035] In one embodiment, the importance coefficient is positively correlated with the Spearman rank correlation coefficient of the risk score, and is negatively correlated with the Pearson correlation coefficient in continuous data, the standardized mutual information in discrete data, and the point biserial correlation coefficient between continuous data and discrete data; the importance coefficient C of the i-th financial data i for:

[0036]

[0037] In the formula, r i represents the Spearman rank correlation coefficient between the ith financial data and its corresponding risk score, α represents the weight factor, ln represents the logarithmic function with base e, r represents the Pearson correlation coefficient in continuous data, NMI represents the standardized mutual information in discrete data, and R represents the point-biserial correlation coefficient between continuous and discrete data.

[0038] For example: if the Spearman rank correlation coefficient between the second debt-to-asset ratio and its risk score is 0.6, the Pearson correlation coefficient in continuous data is 0.4, the standardized mutual information in discrete data is 0.3, the point biserial correlation coefficient between continuous data and discrete data is 0.25, and the weight factor α is 0.5, substitute the above values ​​into the importance coefficient formula for calculation, then the importance coefficient of the debt-to-asset ratio is 0.693.

[0039] S3: Use the importance coefficient to quantify the weight of each classification decision tree in the XGBoost model to obtain a trained XGBoost model.

[0040] In one embodiment, when constructing an XGBoost model, a set of models containing multiple classification decision trees is used, and each decision tree is assigned a weight according to different financial data. In order to improve the accuracy of the model, different weights are assigned to each tree, and the weights of each classification decision tree in the XGBoost model are calculated. The weight is positively correlated with the importance coefficient and negatively correlated with the number of financial data assigned to each classification decision tree. Among them, the weight D of the kth classification decision tree k for:

[0041]

[0042] In the formula, C i represents the importance coefficient of the i-th financial data, and M represents the number of financial data assigned to the k-th classification decision tree.

[0043] For example, if the first classification decision tree is assigned to five financial data, the corresponding importance coefficients are 0.693, 0.55, 0.48, 0.6 and 0.4 respectively, and the weight of the first classification decision tree is calculated as:

[0044]

[0045] Similarly, the weights of other decision trees can be calculated to train the XGBoost model. The training data is the financial data in the above steps and its corresponding importance coefficient. During the training process, the model will adjust its classification effect according to the weight of each classification decision tree to improve the overall classification accuracy, and then obtain a trained XGboost model.

[0046] S4: Use the trained XGboost model to output the final probability of each risk type to which the financial data belongs, and perform risk monitoring on the financial data.

[0047] In this embodiment, the financial data obtained above is input into the trained XGboost model, and the weights of each classification decision tree in the model and the probabilities of each risk type output by each classification decision tree are weighted averaged to obtain the final probability of each risk type to which the financial data belongs, wherein the risk types include: high risk type, medium risk type and low risk type;

[0048] The specific process of risk monitoring is: constructing an XGBoost model, training the XGBoost model, and obtaining a trained XGBoost model;

[0049] Input the financial data into the trained XGBoost model to obtain the final probability of the risk type to which the financial data belongs.

[0050] For example, assuming the risk monitoring result of a piece of financial data: the probability of a high-risk type is 40%, the probability of a medium-risk type is 35%, and the probability of a low-risk type is 25%. The final risk type of the financial data is determined based on the maximum value of the risk probability. Since the probability of a high-risk type is the largest, the financial data is judged to be a high-risk type.

[0051] In this embodiment, the risk types of all financial data can also be automatically monitored, and early warning signals can be issued based on high-risk data. For example, after detecting high-risk data, the system can automatically generate a report and notify relevant responsible personnel to take countermeasures, such as increasing the intensity of review, adjusting credit limits or other security measures, thereby achieving continuous early warning of potential risks.

[0052] The solution of the present invention realizes accurate risk monitoring of supply chain financial data by combining multiple statistical correlation coefficients and an improved XGBoost model. By distinguishing continuous data from discrete data and calculating importance coefficients, the ability to distinguish data features in the model is effectively improved. The improved XGBoost model can perform more accurate risk classification of financial data based on the weights of each classification decision tree and the output risk probability, thereby improving the accuracy and reliability of risk monitoring. This method not only enhances the adaptability of the model to complex data, but also can more effectively identify and warn of potential risks.

[0053] An embodiment of a supply chain financial transaction security early warning system:

[0054] like Figure 2 As shown, a structural block diagram of a supply chain financial transaction security early warning system according to an embodiment of the present invention includes a processor and a memory.

[0055] The present invention also provides a supply chain financial transaction security early warning system. Figure 2 As shown, the system includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a supply chain financial transaction security early warning method according to the above-mentioned invention is implemented.

[0056] The supply chain financial transaction security early warning system also includes other components familiar to those skilled in the art, such as a communication interface, and their settings and functions are known in the art, so they will not be described in detail here.

[0057] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any suitable magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that may be used to store the required information and may be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.

[0058] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0059] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A supply chain financial transaction security early warning method, characterized in that: include: Obtain financial data of the supply chain and its corresponding risk score, and divide the financial data into continuous data and discrete data; Calculating the importance coefficient of the financial data, the importance coefficient is positively correlated with the Spearman rank correlation coefficient of the risk score, and is negatively correlated with the Pearson correlation coefficient in continuous data, the standardized mutual information in discrete data, and the point-biserial correlation coefficient between continuous data and discrete data; Using the improved XGBoost model to perform risk monitoring on the financial data, and obtaining the final probability of each risk type to which the financial data belongs; Among them, the improved XGBoost model includes: calculating the weight of each classification decision tree in the XGBoost model, the weight is positively correlated with the importance coefficient, and negatively correlated with the number of financial data assigned to each classification decision tree; taking a weighted average of the weight of each classification decision tree and the probability of each risk type output by each classification decision tree to obtain the final probability of each risk type to which the financial data belongs, and the risk type corresponding to the maximum value of the final probability is the final risk type of the financial data.

2. A supply chain financial transaction security early warning method according to claim 1, characterized in that: The importance coefficient C of the i-th financial data i for: In the formula, r i represents the Spearman rank correlation coefficient between the ith financial data and its corresponding risk score, α represents the weight factor, ln represents the logarithmic function with base e, r represents the Pearson correlation coefficient in continuous data, NMI represents the standardized mutual information in discrete data, and R represents the point-biserial correlation coefficient between continuous and discrete data.

3. A supply chain financial transaction security early warning method according to claim 1, characterized in that: The weight D of the k-th classification decision tree k for: In the formula, C i represents the importance coefficient of the i-th financial data, and M represents the number of financial data assigned to the k-th classification decision tree.

4. A supply chain financial transaction security early warning method according to claim 1, characterized in that: The continuous data include: debt-to-asset ratio, current ratio, quick ratio, order cycle, transaction amount standard deviation and transaction amount; the discrete data include: number of transactions, number of overdue payments, customer credit score and number of historical defaults.

5. A supply chain financial transaction security early warning method according to claim 1, characterized in that: The specific process of risk monitoring of the financial data is as follows: Constructing an XGBoost model, and training the XGBoost model to obtain a trained XGBoost model; Input the financial data into the trained XGBoost model to obtain the final probability of the risk type to which the financial data belongs.

6. A supply chain financial transaction security early warning method according to claim 1, characterized in that: The risk types include: high risk type, medium risk type and low risk type.

7. A supply chain financial transaction security early warning method according to claim 1, characterized in that: It also includes data preprocessing of financial data using spline interpolation method.

8. A supply chain financial transaction security early warning system, characterized in that: It comprises a memory and a processor, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a supply chain financial transaction security early warning method as described in any one of claims 1 to 7 is implemented.

Images