Financial data processing method and system based on big data analysis

Abstract

The invention discloses a financial data processing method and system based on big data analysis, relates to the technical field of financial data processing, and aims to construct a financial data vector set Fin for a target transaction object, break a data island and provide a comprehensive data basis for multi-dimensional analysis. The income stability is evaluated by calculating a transaction fluctuation index Vt; calculating an income channel authenticity score Ra to evaluate the health of the growth mode; and calculating a potential enumeration value Ld, and converting the operation risk into a specific financial influence. A financial feature vector set Fan is constructed based on the risk indexes, a comprehensive risk index IRI is calculated, a transaction decision AIR is generated and executed by comparing the comprehensive risk index IRI with a decision threshold interval theta, a complex analysis process is concentrated into an objective decision basis, and decision efficiency and scientificity are improved. And finally, iterative optimization is performed according to a historical database HIS, a self-learning and evolution closed loop of the model is established, and long-term accuracy and adaptability are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of financial data processing, and in particular to a financial data processing method and system based on big data analysis. Background Art

[0002] In the modern business environment, financial data processing has expanded far beyond routine bookkeeping and report generation, extending to critical aspects supporting major corporate strategic decision-making. In particular, in complex capital operations such as mergers and acquisitions, investors must conduct in-depth financial due diligence on target companies. This process requires rapid and accurate analysis of vast amounts of financial and non-financial data to assess the target company's true value and identify potential risks. Therefore, a financial data processing method and system capable of integrating and deeply analyzing massive amounts of information is crucial in this application scenario.

[0003] However, in traditional corporate M&A financial due diligence practices, existing data processing methods have significant shortcomings and deficiencies. Background investigation teams are often faced with the challenge of processing massive amounts of data from multiple heterogeneous systems within a very short period of time. Due to technical limitations, analysis relies heavily on manual review and sampling audits by financial experts. This approach is not only inefficient and costly, but also struggles to cover the entire data set and uncover complex risk patterns hidden within massive transaction records. Furthermore, various types of data are often analyzed in isolation, lacking effective correlations between financial data, business data, and production and operations data. This results in superficial analysis conclusions and an inability to form a complete risk profile.

[0004] The aforementioned current situation and shortcomings are primarily caused by the conflict between data volume and analytical tools. Under pressure from tight transaction deadlines, analytical teams are forced to compromise with sampling audits, which directly leads to a variety of anomalies. First, it makes it difficult to detect carefully concealed financial fraud or structural risks, such as fabricated transactions through specific business channels to inflate revenue, or potential financial drag signaled by persistent equipment failures that remains unrecognized in the financial statements. Second, this one-sided risk assessment can seriously mislead transaction valuation models, potentially causing the acquirer to pay a price far exceeding the target's actual value. Ultimately, when these undetected risks materialize after the transaction is completed, the actual profitability of the acquired company can fall far short of pre-transaction expectations, resulting in significant losses or even failure of the acquisition. Summary of the Invention

[0005] In view of the deficiencies of the prior art, the present invention provides a financial data processing method and system based on big data analysis, which solves the problems mentioned in the background technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: A financial data processing method based on big data analysis, comprising the following steps:

[0007] S1. For the target transaction object, collect account transaction data Dtr, business operation data Dre, and production operation data Dop through the API interface to construct the financial data vector set Fin;

[0008] S2. Based on the account transaction data Dtr in the financial feature vector set Fin, a transaction volatility assessment algorithm is constructed to calculate the transaction volatility index Vt to assess the stability and health of income;

[0009] S3. Based on the business operation data Dre in the financial feature vector set Fin, a revenue channel authenticity evaluation algorithm is constructed to calculate the revenue channel authenticity score Ra and evaluate the stability and health of the revenue growth model.

[0010] S4. Based on the production operation data Dop in the financial feature vector set Fin, a potential drag value assessment algorithm is constructed to calculate the potential drag value Ld caused by equipment failure and product returns;

[0011] S5. Construct a financial feature vector set Fan based on the transaction volatility index Vt, the revenue channel authenticity score Ra, and the potential drag value Ld. Calculate the comprehensive risk index IRI and compare it with the decision threshold interval θ. Generate a transaction decision AIR based on the comparison results and execute it.

[0012] S6. After executing the transaction decision AIR, the financial data vector set Fin and the actual profit amount APA of the target transaction object are continuously collected and the transaction decision accuracy rate AIR_true is analyzed. Iterative optimization is performed based on the transaction decision accuracy rate AIR_true, and the financial feature vector set Fin, the financial feature vector set Fan and the comprehensive risk index IRI are stored in the historical database HIS.

[0013] Preferably, S1 includes S11 and S12;

[0014] S1. For the target transaction object, collect account transaction data Dtr, business operation data Dre, and production operation data Dop through the API interface to construct the financial data vector set Fin;

[0015] Among them, the accounting transaction data Dtr is collected from the ERP system of the target transaction object through the API interface, the business operation data Dre is collected from the CRM system of the target transaction object through the API interface, and the production operation data Dop is collected from the MES system of the target transaction object through the API interface to construct the financial data vector set Fin.

[0016] Preferably, S12, the account transaction data Dtr includes the total number of customers C of the target transaction object, the customer identifier c, the number of transactions nc of the c-th customer, the total number of transactions N and the standard deviation σAt of the single transaction amount;

[0017] The business operation data Dre includes the total number of transaction channels M of the target transaction object, the channel identifier m, the time span Δtm from the establishment of the mth channel to the end of the observation period, the actual total revenue rm of the mth channel during the observation period, the expected revenue E[rm] of the mth channel calculated based on the historical database HIS, and the life cycle decay constant km of the mth channel;

[0018] The production operation data Dop includes the average return rate Tp of the target transaction object within the preset statistical period T, the standard deviation of the return rate δTp within the preset statistical period T, the total number of sales units Nops, the number of production equipment Z, the equipment identifier z and the average annual failure rate λz of the zth equipment.

[0019] Preferably, S2 includes S21;

[0020] S21. Based on the account transaction data Dtr in the financial feature vector set Fin, a transaction volatility assessment algorithm is constructed. A customer breadth correction factor is constructed based on the total number of customers C. Based on the customer identifier c and the total number of transactions N, and incorporating the Herfindahl-Hirschman index concept, the square of the ratio of the number of transactions nc of the cth customer to the total number of transactions N is calculated and the square root is taken. Finally, the standard deviation of the single transaction amount σAt is used as the transaction amount volatility factor to calculate the transaction volatility index Vt.

[0021] The trading volatility evaluation algorithm expression is as follows:

[0022] ;

[0023] In the formula, ∑ represents the summation operation, Represents the square root operation.

[0024] Preferably, S3 includes S31;

[0025] S31. Based on the business operation data Dre in the financial feature vector set Fin, construct a revenue channel authenticity assessment algorithm. Calculate the ratio of the actual total revenue rm of the mth channel during the observation period to the expected revenue E[rm] of the mth channel, perform an absolute value operation on the logarithm, and calculate the negative product of the time span Δtm from the establishment of the mth channel to the end of the observation period and the life cycle decay constant km of the mth channel. Input this into the natural exponential function exp. Combined with the total number of transaction channels M, calculate the revenue channel authenticity score Ra.

[0026] The income channel authenticity evaluation algorithm expression is as follows:

[0027] ;

[0028] In the formula, |·| represents the absolute value operation, ln represents the logarithm operation with the natural constant e as the base, exp represents the natural exponential function, and ε represents the regularization term to prevent the denominator from being zero, specifically 1*10 -8 .

[0029] Preferably, S4 includes S41;

[0030] S41. Based on the production and operation data Dop in the financial feature vector set Fin, a potential drag value assessment algorithm is constructed. A normal distribution model is constructed using the average return rate Tp and the standard deviation of the return rate δTp within a preset statistical period T as parameters, and the result is multiplied by the single return cost src and the total number of units sold Nops. A Poisson process model is applied to the annual average failure rate λz of the zth device and the preset statistical period T, and the result is multiplied by the preset economic loss Fz for a single failure of the zth device. The potential drag value Ld is calculated by combining the normal distribution model terms with the Poisson process model terms.

[0031] Among them, the potential drag value evaluation algorithm expression is as follows:

[0032] ;

[0033] In the formula, src represents the average cost of a single return preset by the cost accounting expert, p0 represents the preset benchmark return rate, and x represents the integral variable, which specifically means the future return rate independent variable in the interval [p0, 1]. represents the square root operation, It represents the total probability of predicting the future return rate x within the interval [p0, 1] using integral calculation. π represents the ratio of pi, which is a double-precision floating-point number with 16 significant digits. Fz represents the preset economic loss of a single failure of the zth device obtained from analysis of the historical database HIS. exp represents the natural exponential function.

[0034] Preferably, S5 includes S51 and S52;

[0035] S51. Construct a financial feature vector set Fan based on the transaction volatility index Vt, the revenue channel authenticity score Ra, and the potential drag value Ld, and construct a comprehensive risk assessment algorithm. Standardize the transaction volatility index Vt and the revenue channel authenticity score Ra using the transaction volatility index reference value Vref and the revenue channel authenticity score reference value Rref. Compare the potential drag value Ld with the one-year EBITDA G of the transaction target. After eliminating dimensional differences, perform a weighted addition, and input the weighted summation result into the hyperbolic tangent function tanh to calculate the comprehensive risk index IRI.

[0036] Among them, by analyzing the transaction volatility index Vt_HIS and the income channel authenticity score Ra_HIS in the historical database HIS, the transaction volatility index reference value Vref and the income channel authenticity score reference value Rref are obtained. The comprehensive risk assessment algorithm expression is as follows:

[0037] ;

[0038] Wherein, α represents the influencing factor of the transaction volatility index Vt obtained by performing multiple regression analysis on the historical database HIS, β represents the influencing factor of the income channel authenticity score Ra obtained by performing multiple regression analysis on the historical database HIS, and γ represents the influencing factor of the potential drag value Ld obtained by performing multiple regression analysis on the historical database HIS.

[0039] Preferably, S52, comparing the comprehensive risk index IRI with the decision threshold interval θ, generating a transaction decision AIR based on the comparison result and executing the transaction decision AIR;

[0040] Among them, the decision threshold interval θ includes the first-level decision threshold θ1 and the second-level decision threshold θ2;

[0041] If the comprehensive risk index IRI is less than the first-level decision threshold θ1, the target transaction object is determined to meet the transaction requirements, and the transaction decision AIR is to trade with the target transaction object in accordance with the agreed terms;

[0042] If the first-level decision threshold θ1 ≤ comprehensive risk index IRI < second-level decision threshold θ2, the target transaction object is judged to have transaction risks but meet the transaction requirements. The transaction decision AIR is to renegotiate with the transaction target, request a reduction in the acquisition valuation and include a seller guarantee clause in the acquisition agreement.

[0043] If the first-level decision threshold θ1 ≥ the second-level decision threshold θ2, it is determined that the target transaction object has transaction risks and does not meet the transaction requirements, and the transaction decision AIR is to give up the transaction with the target transaction object.

[0044] Preferably, S6 includes S61;

[0045] S61. After the transaction decision AIR is executed, the financial data vector set Fin of the target transaction object within one year is continuously collected and the transaction decision accuracy rate AIR_true is analyzed;

[0046] If the actual profit APA of the target transaction object in the transaction within one year is greater than the preset expected profit EPA, or the actual profit APA of the target transaction object in the transaction that is abandoned within one year is less than the preset expected profit EPA, then the transaction decision is recorded as correct;

[0047] If the actual profit APA of the target transaction object within one year is less than the preset expected profit EPA, or the actual profit APA of the target transaction object within one year is greater than the preset expected profit EPA, then it will be recorded as a transaction decision error false;

[0048] Calculate the trading decision accuracy rate AIR_true based on the number of correct trading decisions (true) and the number of incorrect trading decisions (false). Analyze the trading decision accuracy rate AIR_true every quarter.

[0049] If the transaction decision accuracy rate AIR_true ≥ the preset expected transaction decision accuracy rate E_true, then there is no need for iterative optimization;

[0050] If the transaction decision accuracy rate AIR_true is less than the preset expected transaction decision accuracy rate E_true, iterative optimization is required;

[0051] Professionals in this field adjust the influencing factor α of the transaction volatility index Vt, the influencing factor β of the income channel authenticity score Ra, and the influencing factor γ of the potential drag value Ld in the comprehensive risk assessment algorithm, and conduct a multivariate regression analysis based on the historical database HIS and the transaction decision accuracy rate AIR_true. The adjusted influencing factor α of the transaction volatility index Vt, the influencing factor β of the income channel authenticity score Ra, and the influencing factor γ of the potential drag value Ld are calibrated again, and the financial feature vector set Fin, the financial feature vector set Fan, and the comprehensive risk index IRI are stored in the historical database HIS.

[0052] A financial data processing system based on big data analysis, including a financial data acquisition module, a transaction fluctuation analysis module, a revenue channel authenticity analysis module, a potential drag value analysis module, a comprehensive risk index analysis and evaluation module, and an iterative optimization module;

[0053] The data collection module collects accounting transaction data Dtr, business operation data Dre, and production operation data Dop from the target transaction object through the API interface to construct the financial data vector set Fin;

[0054] The transaction volatility analysis module constructs a transaction volatility assessment algorithm based on the account transaction data Dtr in the financial feature vector set Fin, calculates the transaction volatility index Vt, and evaluates the stability and health of income;

[0055] The revenue channel authenticity analysis module constructs a revenue channel authenticity evaluation algorithm based on the business operation data Dre in the financial feature vector set Fin, calculates the revenue channel authenticity score Ra, and evaluates the stability and health of the revenue growth model;

[0056] The potential drag value analysis module constructs a potential drag value assessment algorithm based on the production operation data Dop in the financial feature vector set Fin to calculate the potential drag value Ld caused by equipment failure and product returns;

[0057] The comprehensive risk index analysis and assessment module constructs a financial feature vector set Fan based on the transaction volatility index Vt, the revenue channel authenticity score Ra, and the potential drag value Ld. It then calculates the comprehensive risk index IRI and compares it with the decision threshold interval θ. Based on the comparison results, it generates and executes the transaction decision AIR.

[0058] After the iterative optimization module completes the transaction decision AIR, it continuously collects the financial data vector set Fin and actual profit amount APA of the target transaction object and analyzes the transaction decision accuracy AIR_true. It performs iterative optimization based on the transaction decision accuracy AIR_true and stores the financial feature vector set Fin, financial feature vector set Fan and comprehensive risk index IRI in the historical database HIS.

[0059] The present invention provides a financial data processing method and system based on big data analysis, which has the following beneficial effects:

[0060] (1) Through the API interface, the isolated accounting transaction data Dtr, business operation data Dre, and production operation data Dop of the target transaction object are fully collected and deeply integrated to construct a unified financial data vector set Fin. On this solid data foundation, the system deploys a series of interlocking evaluation algorithms, aiming to penetrate the surface of financial statements and deeply analyze their intrinsic quality. It uses specially constructed algorithms to quantify the stability of transaction behavior, the authenticity of revenue sources, and the potential financial drag at the operational level, generating a transaction volatility index Vt, a revenue channel authenticity score Ra, and a potential drag value Ld. Ultimately, these multi-dimensional risk indicators are intelligently aggregated into a comprehensive decision-making basis, and through an iterative optimization mechanism capable of self-learning, it ensures that the analysis model can continuously learn from historical cases and continuously improve its accuracy and adaptability, thereby providing a dynamic, quantitative, and reliable technical support for complex financial decisions.

[0061] (2) For the target transaction object, the accounting transaction data Dtr in the ERP system, the business operation data Dre in the CRM system, and the production operation data Dop in the MES system are collected through the API interface, and a unified financial data vector set Fin is constructed. This breaks the data barriers between finance, business, and operations in traditional analysis, laying a solid and comprehensive data foundation for subsequent cross-domain and in-depth correlation risk analysis. Then, by constructing a transaction volatility assessment algorithm, the transaction volatility index Vt is calculated, which can quantitatively assess the stability of revenue and customer concentration risk. On this basis, the depth of analysis is further improved. By constructing a revenue channel authenticity assessment algorithm to calculate the revenue channel authenticity score Ra, it effectively identifies the false prosperity created by means such as channel pressure, and penetrates to assess the true health of the revenue growth model. Furthermore, by constructing a potential drag value assessment algorithm to calculate the potential drag value Ld, operational risks such as equipment failure and product returns that have not yet been reflected in the financial statements are prospectively quantified as specific future financial losses, providing a key, future-oriented risk deduction basis for valuation.

[0062] (3) Based on the transaction volatility index Vt, revenue channel authenticity score Ra, and potential drag value Ld calculated in the previous step, a comprehensive risk assessment algorithm is used to aggregate them into a single comprehensive risk index IRI. This transforms the complex, multi-dimensional analysis results into an intuitive and comparable decision-making basis. By comparing them with the decision threshold interval θ, a clear transaction decision AIR can be automatically generated, greatly improving the objectivity, scientificity, and efficiency of the decision. Next, a key feedback and optimization closed loop is constructed. After executing the transaction decision AIR, the actual operating results will be continuously tracked, and the model will be iteratively optimized based on the historical database HIS. This gives the entire analysis method the ability to self-learn and adapt, and it can learn from past successes and failures and continuously calibrate its core algorithm parameters, making future analysis and predictions more accurate, and ensuring and improving the accuracy of transaction decisions AIR_true in the long term. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 This is a schematic diagram of the steps of a financial data processing method based on big data analysis of the present invention;

[0064] Figure 2 This is a schematic diagram of a financial data processing system based on big data analysis according to the present invention;

[0065] Figure 3 Flowchart of data processing for generating transaction decision AIR. DETAILED DESCRIPTION

[0066] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0067] Example 1

[0068] The present invention provides a financial data processing method based on big data analysis, please refer to Figure 1 , including the following steps:

[0069] S1. For the target transaction object, collect account transaction data Dtr, business operation data Dre, and production operation data Dop through the API interface to construct the financial data vector set Fin;

[0070] S2. Based on the account transaction data Dtr in the financial feature vector set Fin, a transaction volatility assessment algorithm is constructed to calculate the transaction volatility index Vt to assess the stability and health of income;

[0071] S3. Based on the business operation data Dre in the financial feature vector set Fin, a revenue channel authenticity evaluation algorithm is constructed to calculate the revenue channel authenticity score Ra and evaluate the stability and health of the revenue growth model.

[0072] S4. Based on the production operation data Dop in the financial feature vector set Fin, a potential drag value assessment algorithm is constructed to calculate the potential drag value Ld caused by equipment failure and product returns;

[0073] S5. Construct a financial feature vector set Fan based on the transaction volatility index Vt, the revenue channel authenticity score Ra, and the potential drag value Ld. Calculate the comprehensive risk index IRI and compare it with the decision threshold interval θ. Generate a transaction decision AIR based on the comparison results and execute it.

[0074] S6. After executing the transaction decision AIR, the financial data vector set Fin and the actual profit amount APA of the target transaction object are continuously collected and the transaction decision accuracy rate AIR_true is analyzed. Iterative optimization is performed based on the transaction decision accuracy rate AIR_true, and the financial feature vector set Fin, the financial feature vector set Fan and the comprehensive risk index IRI are stored in the historical database HIS.

[0075] This embodiment overcomes the limitations of traditional due diligence, which relies on sample audits and fragmented data sources, by comprehensively collecting accounting transaction data Dtr, business operation data Dre, and production operation data Dop and constructing a unified financial data vector set Fin. Furthermore, by calculating the transaction volatility index Vt, the revenue channel authenticity score Ra, and the potential drag value Ld, the analysis penetrates beyond surface financial figures to deeper dimensions such as transaction quality, revenue source authenticity, and future operating costs. This accurately reveals carefully concealed structural risks and potential liabilities that are easily overlooked using traditional methods. Furthermore, by aggregating these multi-dimensional risks into a single comprehensive risk index (IRI) and generating a clear transaction decision AIR, the judgment process, once based on subjective experience, is transformed into an objective, quantifiable, and reproducible scientific decision-making process. This significantly improves the accuracy of transaction valuations and effectively avoids the risk of excessive acquisition costs due to information asymmetry. Finally, by introducing the closed-loop mechanism of iterative optimization of the historical database HIS, the long-term effectiveness and self-evolution capability of the analysis algorithm are ensured, which can continuously improve the accuracy of transaction decisions AIR_true and fundamentally reduce the possibility of post-investment integration failure due to incorrect evaluation.

[0076] Example 2

[0077] This embodiment is explained in Example 1, please refer to Figure 1 and Figure 3,Specifically: S1 includes S11 and S12;

[0078] S1. For the target transaction object, collect account transaction data Dtr, business operation data Dre, and production operation data Dop through the API interface to construct the financial data vector set Fin;

[0079] Among them, the accounting transaction data Dtr is collected from the ERP system of the target transaction object through the API interface, the business operation data Dre is collected from the CRM system of the target transaction object through the API interface, and the production operation data Dop is collected from the MES system of the target transaction object through the API interface to construct the financial data vector set Fin;

[0080] S12. Account transaction data Dtr includes the total number of target transaction customers C, customer identifier c, number of transactions of the c-th customer nc, total number of transactions N, and standard deviation of the single transaction amount σAt;

[0081] The business operation data Dre includes the total number of transaction channels M of the target transaction object, the channel identifier m, the time span Δtm from the establishment of the mth channel to the end of the observation period, the actual total revenue rm of the mth channel during the observation period, the expected revenue E[rm] of the mth channel calculated based on the historical database HIS, and the life cycle decay constant km of the mth channel;

[0082] The production operation data Dop includes the average return rate Tp of the target transaction object within the preset statistical period T, the standard deviation of the return rate δTp within the preset statistical period T, the total number of sales units Nops, the number of production equipment Z, the equipment identifier z, and the average annual failure rate λz of the z-th equipment;

[0083] S2 includes S21;

[0084] S21. Based on the account transaction data Dtr in the financial feature vector set Fin, a transaction volatility assessment algorithm is constructed. A customer breadth correction factor is constructed based on the total number of customers C. Based on the customer identifier c and the total number of transactions N, and incorporating the Herfindahl-Hirschman index concept, the square of the ratio of the number of transactions nc of the cth customer to the total number of transactions N is calculated and the square root is taken. Finally, the standard deviation of the single transaction amount σAt is used as the transaction amount volatility factor to calculate the transaction volatility index Vt.

[0085] The trading volatility evaluation algorithm expression is as follows:

[0086] ;

[0087] In the formula, ∑ represents the summation operation, Indicates square root operation;

[0088] S3 includes S31;

[0089] S31. Based on the business operation data Dre in the financial feature vector set Fin, construct a revenue channel authenticity assessment algorithm. Calculate the ratio of the actual total revenue rm of the mth channel during the observation period to the expected revenue E[rm] of the mth channel, perform an absolute value operation on the logarithm, and calculate the negative product of the time span Δtm from the establishment of the mth channel to the end of the observation period and the life cycle decay constant km of the mth channel. Input this into the natural exponential function exp. Combined with the total number of transaction channels M, calculate the revenue channel authenticity score Ra.

[0090] The income channel authenticity evaluation algorithm expression is as follows:

[0091] ;

[0092] In the formula, |·| represents the absolute value operation, ln represents the logarithm operation with the natural constant e as the base, exp represents the natural exponential function, and ε represents the regularization term to prevent the denominator from being zero, specifically 1*10 -8 ;

[0093] The purpose of this formula is to construct a surprise factor that quantifies the relative degree of deviation from expectations by calculating the ratio of the actual total revenue rm of the mth channel during the observation period to the expected revenue E[rm] of the mth channel and performing an absolute value operation on the logarithm. , quantifies the abnormal degree of income of the mth channel, and constructs the maturity adjustment factor by calculating the negative product of the time span Δtm from the establishment of the mth channel to the end of the observation period and the life cycle decay constant km of the mth channel, and inputting it into the natural exponential function exp to form an exponential decay function For new channels, the time span Δtm from the establishment of the mth channel to the end of the observation period tends to 0, and the maturity adjustment factor approaches 1, which means that the mth channel will be fully included in the penalty. For old channels, the time span Δtm from the establishment of the mth channel to the end of the observation period will continue to increase, and the maturity adjustment factor approaches 0, which will greatly weaken the penalty brought by non-mth channels. Finally, combining the surprise factor, maturity adjustment factor and the total number of transaction channels M, the revenue channel authenticity score Ra is calculated to quantify the revenue channel quality and growth authenticity of the target transaction object;

[0094] S4 includes S41;

[0095] S41. Based on the production and operation data Dop in the financial feature vector set Fin, a potential drag value assessment algorithm is constructed. A normal distribution model is constructed using the average return rate Tp and the standard deviation of the return rate δTp within a preset statistical period T as parameters, and the result is multiplied by the single return cost src and the total number of units sold Nops. A Poisson process model is applied to the annual average failure rate λz of the zth device and the preset statistical period T, and the result is multiplied by the preset economic loss Fz for a single failure of the zth device. The potential drag value Ld is calculated by combining the normal distribution model terms with the Poisson process model terms.

[0096] Among them, the potential drag value evaluation algorithm expression is as follows:

[0097] ;

[0098] In the formula, src represents the average cost of a single return preset by cost accounting experts, p0 represents the preset benchmark return rate, and x represents the integral independent variable, which specifically means the future return rate in the interval [p0, 1]. represents the square root operation, Indicates the total probability of the integral variable x falling within the interval [p0, 1] using integral operation. π represents the ratio of pi, which is a double-precision floating-point number with 16 significant digits. Fz represents the preset economic loss of a single failure of the zth device obtained from the analysis of the historical database HIS. exp represents the natural exponential function.

[0099] The purpose of this formula is to construct a normal distribution model with the average return rate Tp and the standard deviation of the return rate δTp within the preset statistical period T as parameters, and perform definite integral operations on the interval exceeding the acceptable benchmark to obtain a cumulative probability of a deterioration in the return rate in the future. Then, the expected loss of product returns is calculated by multiplying the cumulative probability of the return rate deteriorating in the future by the maximum potential loss consisting of the single return cost src and the total number of sales Nops. By applying the Poisson process model to the zth device, the probability that the zth device will fail at least once in the next year is calculated based on the annual average failure rate λz of the zth device and the preset statistical period T, and multiplying it with the preset economic loss Fz of a single failure of the zth device to obtain the expected failure loss of a single device. And perform cumulative calculations to obtain the expected failure losses of all devices Finally, the expected loss from product returns and the expected loss from failure of all equipment are used to calculate the potential drag value Ld, which quantifies the potential economic losses to the company in the next year due to operational efficiency and quality issues.

[0100] In this embodiment, by comprehensively collecting accounting transaction data Dtr, business operation data Dre, and production operation data Dop and constructing a unified financial data vector set Fin, the limitations of traditional due diligence, which relies on sampling audits and fragmented data sources, are overcome. Next, abstract financial risks are transformed into measurable and interpretable quantitative indicators, achieving a significant improvement in analytical depth. Specifically, the uniqueness of the transaction volatility assessment algorithm lies in that it does not view transaction amount or number of customers in isolation. Instead, it multiplies the standard deviation σAt of the single transaction amount, which measures the degree of dispersion of the amount, with the customer transaction frequency concentration factor, which draws on the concept of the Herfindahl-Hirschman Index. This simultaneously captures two different types of risks—the "instability" of transaction scale and the "dependence" of customer sources—and quantifies the compound risk of their superposition. The resulting assessment results are far more profound and comprehensive than single-dimensional analysis. The revenue channel authenticity assessment algorithm constructs a surprise factor based on the ratio of the actual total revenue rm of the mth channel during the observation period to the expected revenue E[rm] of the mth channel to identify deviations from historical patterns. It also incorporates the time span Δtm from the establishment of the mth channel to the end of the observation period and an exponential decay function of the life cycle constant km as maturity adjustment factors. This intelligently distinguishes between normal fluctuations in mature channels and highly suspicious explosive growth in emerging channels, accurately removing "water" and restoring the true quality of revenue. The beneficial effect of the potential drag value assessment algorithm lies in its forward-looking risk monetization capabilities. Based on production and operation data (Dop), it proactively applies probabilistic models based on normal distribution and Poisson processes to calculate the probability-weighted expected value of possible future returns and equipment failure losses that exceed standard benchmarks. This transforms fuzzy, non-financial operational risks into concrete financial values—potential drag values ​​Ld—that can be directly deducted in the valuation model, providing a robust quantitative basis for decision-making.

[0101] Example 3

[0102] This embodiment is explained in Example 2, please refer to Figure 1 and Figure 3 ,Specifically: S5 includes S51 and S52;

[0103] S51. Construct a financial feature vector set Fan based on the transaction volatility index Vt, the revenue channel authenticity score Ra, and the potential drag value Ld, and construct a comprehensive risk assessment algorithm. Standardize the transaction volatility index Vt and the revenue channel authenticity score Ra using the transaction volatility index reference value Vref and the revenue channel authenticity score reference value Rref. Compare the potential drag value Ld with the one-year EBITDA G of the transaction target. After eliminating dimensional differences, perform a weighted addition, and input the weighted summation result into the hyperbolic tangent function tanh to calculate the comprehensive risk index IRI.

[0104] Among them, by analyzing the transaction volatility index Vt_HIS and the income channel authenticity score Ra_HIS in the historical database HIS, the transaction volatility index reference value Vref and the income channel authenticity score reference value Rref are obtained. The comprehensive risk assessment algorithm expression is as follows:

[0105] ;

[0106] Where α represents the influencing factor of the transaction volatility index Vt obtained by performing a multiple regression analysis on the historical database HIS, β represents the influencing factor of the revenue channel authenticity score Ra obtained by performing a multiple regression analysis on the historical database HIS, and γ represents the influencing factor of the potential drag value Ld obtained by performing a multiple regression analysis on the historical database HIS.

[0107] S52. Compare the comprehensive risk index IRI with the decision threshold interval θ, generate a transaction decision AIR based on the comparison result, and execute the transaction decision;

[0108] Among them, the decision threshold interval θ includes the first-level decision threshold θ1 and the second-level decision threshold θ2;

[0109] If the comprehensive risk index IRI is less than the first-level decision threshold θ1, the target transaction object is determined to meet the transaction requirements, and the transaction decision AIR is to trade with the target transaction object in accordance with the agreed terms;

[0110] If the first-level decision threshold θ1 ≤ comprehensive risk index IRI < second-level decision threshold θ2, the target transaction object is judged to have transaction risks but meet the transaction requirements. The transaction decision AIR is to renegotiate with the transaction target, request a reduction in the acquisition valuation and include a seller guarantee clause in the acquisition agreement.

[0111] If the first-level decision threshold θ1 ≥ the second-level decision threshold θ2, the target transaction object is judged to have transaction risks and does not meet the transaction requirements, and the transaction decision AIR is to give up the transaction with the target transaction object;

[0112] The specific calculation example of the comprehensive risk index IRI and the generation example of the trading decision AIR are as follows:

[0113] Trading volatility index Vt=150.0, trading volatility index reference value Vref=400.0;

[0114] The income channel authenticity score Ra=0.97, the income channel authenticity score reference value Rref=0.90;

[0115] Potential drag value Ld = 6,000,000 yuan, EBITDA G = 60,000,000 yuan;

[0116] The impact factor of the trading volatility index Vt is α=0.4;

[0117] The impact factor of the income channel authenticity score Ra is β=0.8;

[0118] The impact factor of potential drag value Ld is γ=1.0;

[0119] The first-level decision threshold θ1=0.2, the second-level decision threshold θ2=0.6;

[0120] The specific calculation of the comprehensive risk index IRI is as follows:

[0121] ;

[0122] The first-level decision threshold θ1 is less than the comprehensive risk index IRI and less than the second-level decision threshold θ2. The target transaction object has transaction risks but meets the transaction requirements. The transaction decision AIR is to renegotiate with the transaction target, requiring a reduction in the acquisition valuation and the inclusion of a seller guarantee clause in the acquisition agreement.

[0123] In this embodiment, the comprehensive risk assessment algorithm standardizes and normalizes each sub-item of risk by introducing a reference value for the transaction volatility index Vref, a reference value for the revenue channel authenticity score Rref, and a target EBITDA G. The benefit of this design is that it allows for the comparison of risk indicators of varying dimensions and nature within a fair and business-relevant framework. Furthermore, by weighting the impact factor α of the transaction volatility index Vt, the impact factor β of the revenue channel authenticity score Ra, and the impact factor γ of the potential drag value Ld, derived from regression analysis of the historical database HIS, this ensures that risk aggregation is based on historical experience rather than subjective assumptions. The resulting comprehensive risk index (IRI), outputted via the hyperbolic tangent function (tanh), is therefore highly objective and comparable. Furthermore, by establishing a decision threshold interval θ consisting of a primary decision threshold θ1 and a secondary decision threshold θ2, the algorithm transcends the traditional "pass / fail" binary judgment. This multi-level decision-making framework intelligently generates different levels of transaction decisions (AIR) based on the different ranges of the comprehensive risk index (IRI), ranging from "transacting in accordance with agreed terms" to "requesting a valuation reduction and adding seller guarantees" to "abandoning the transaction." It provides refined, data-driven strategic support for complex commercial negotiations, making decision-making a precise science rather than a vague art, greatly enhancing investors' initiative and certainty in transactions.

[0124] Example 4

[0125] This embodiment is explained in Example 3, please refer to Figure 1, specifically: S6 includes S61;

[0126] S61. After the transaction decision AIR is executed, the financial data vector set Fin of the target transaction object within one year is continuously collected and the transaction decision accuracy rate AIR_true is analyzed;

[0127] If the actual profit APA of the target transaction object in the transaction within one year is greater than the preset expected profit EPA, or the actual profit APA of the target transaction object in the transaction that is abandoned within one year is less than the preset expected profit EPA, then the transaction decision is recorded as correct;

[0128] If the actual profit APA of the target transaction object within one year is less than the preset expected profit EPA, or the actual profit APA of the target transaction object within one year is greater than the preset expected profit EPA, then it will be recorded as a transaction decision error false;

[0129] Calculate the trading decision accuracy rate AIR_true based on the number of correct trading decisions (true) and the number of incorrect trading decisions (false). Analyze the trading decision accuracy rate AIR_true every quarter.

[0130] If the transaction decision accuracy rate AIR_true ≥ the preset expected transaction decision accuracy rate E_true, then there is no need for iterative optimization;

[0131] If the transaction decision accuracy rate AIR_true is less than the preset expected transaction decision accuracy rate E_true, iterative optimization is required;

[0132] Professionals in this field adjust the influencing factor α of the transaction volatility index Vt, the influencing factor β of the income channel authenticity score Ra, and the influencing factor γ of the potential drag value Ld in the comprehensive risk assessment algorithm, and conduct a multivariate regression analysis based on the historical database HIS and the transaction decision accuracy rate AIR_true. The adjusted influencing factor α of the transaction volatility index Vt, the influencing factor β of the income channel authenticity score Ra, and the influencing factor γ of the potential drag value Ld are calibrated again, and the financial feature vector set Fin, the financial feature vector set Fan, and the comprehensive risk index IRI are stored in the historical database HIS.

[0133] In this embodiment, by constructing a closed-loop iterative mechanism based on feedback from real business results, the entire analysis system is equipped with dynamic self-learning and evolution capabilities, thereby ensuring the long-term effectiveness and accuracy of the algorithm. Its special feature is that it first establishes a clear and quantifiable standard of right and wrong for each transaction decision AIR by comparing the actual profit amount APA with the expected profit amount EPA. The fuzzy "investment success or failure" is converted into a machine-understandable, black-and-white label of "transaction decision correct true" or "transaction decision wrong false", providing high-quality, unambiguous training data for the subsequent machine learning process, which is the fundamental prerequisite for achieving intelligent optimization. More importantly, a set of automated model evolution processes is constructed. By calculating the transaction decision accuracy rate AIR_true and comparing it with the expected transaction decision accuracy rate E_true, the system can intelligently judge whether the performance of its own model has declined and trigger the optimization program in a timely manner. By conducting a new round of multivariate regression analysis on all cases in the historical database (HIS), the system automatically recalibrates the influencing factors α of the transaction volatility index Vt, β of the revenue channel authenticity score Ra, and γ of the potential drag value Ld in the comprehensive risk assessment algorithm. This means the algorithm can learn from experience and dynamically adjust its emphasis on different risk dimensions, enabling its risk assessment model to keep pace with market changes and reflect the evolving risk preferences of investment teams. Ultimately, this represents a fundamental leap from a static expert rule-based system to a dynamic, intelligent system that continuously accumulates experience.

[0134] Example 5

[0135] A financial data processing system based on big data analysis, please refer to Figure 2 Specifically, it includes financial data collection module, transaction fluctuation analysis module, revenue channel authenticity analysis module, potential drag value analysis module, comprehensive risk index analysis and evaluation module, and iterative optimization module;

[0136] The data collection module collects accounting transaction data Dtr, business operation data Dre, and production operation data Dop from the target transaction object through the API interface to construct the financial data vector set Fin;

[0137] The transaction volatility analysis module constructs a transaction volatility assessment algorithm based on the account transaction data Dtr in the financial feature vector set Fin, calculates the transaction volatility index Vt, and evaluates the stability and health of income;

[0138] The revenue channel authenticity analysis module constructs a revenue channel authenticity evaluation algorithm based on the business operation data Dre in the financial feature vector set Fin, calculates the revenue channel authenticity score Ra, and evaluates the stability and health of the revenue growth model;

[0139] The potential drag value analysis module constructs a potential drag value assessment algorithm based on the production operation data Dop in the financial feature vector set Fin to calculate the potential drag value Ld caused by equipment failure and product returns;

[0140] The comprehensive risk index analysis and assessment module constructs a financial feature vector set Fan based on the transaction volatility index Vt, the revenue channel authenticity score Ra, and the potential drag value Ld. It then calculates the comprehensive risk index IRI and compares it with the decision threshold interval θ. Based on the comparison results, it generates and executes the transaction decision AIR.

[0141] After the iterative optimization module completes the transaction decision AIR, it continuously collects the financial data vector set Fin and actual profit amount APA of the target transaction object and analyzes the transaction decision accuracy AIR_true. It performs iterative optimization based on the transaction decision accuracy AIR_true and stores the financial feature vector set Fin, financial feature vector set Fan and comprehensive risk index IRI in the historical database HIS.

[0142] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A financial data processing method based on big data analysis, characterized by: The following steps are involved: S1. For the target transaction object, collect account transaction data Dtr, business operation data Dre, and production operation data Dop through the API interface to construct the financial data vector set Fin; S2. Based on the account transaction data Dtr in the financial feature vector set Fin, a transaction volatility assessment algorithm is constructed to calculate the transaction volatility index Vt to assess the stability and health of income; S3. Based on the business operation data Dre in the financial feature vector set Fin, a revenue channel authenticity evaluation algorithm is constructed to calculate the revenue channel authenticity score Ra and evaluate the stability and health of the revenue growth model. S4. Based on the production operation data Dop in the financial feature vector set Fin, a potential drag value assessment algorithm is constructed to calculate the potential drag value Ld caused by equipment failure and product returns; S5. Construct a financial feature vector set Fan based on the transaction volatility index Vt, the revenue channel authenticity score Ra, and the potential drag value Ld. Calculate the comprehensive risk index IRI and compare it with the decision threshold interval θ. Generate a transaction decision AIR based on the comparison results and execute it. S6. After executing the transaction decision AIR, the financial data vector set Fin and the actual profit amount APA of the target transaction object are continuously collected and the transaction decision accuracy rate AIR_true is analyzed. Iterative optimization is performed based on the transaction decision accuracy rate AIR_true, and the financial feature vector set Fin, the financial feature vector set Fan and the comprehensive risk index IRI are stored in the historical database HIS.

2. The financial data processing method based on big data analysis according to claim 1, characterized in that: S1 includes S11 and S12; S1. For the target transaction object, collect account transaction data Dtr, business operation data Dre, and production operation data Dop through the API interface to construct the financial data vector set Fin; Among them, the accounting transaction data Dtr is collected from the ERP system of the target transaction object through the API interface, the business operation data Dre is collected from the CRM system of the target transaction object through the API interface, and the production operation data Dop is collected from the MES system of the target transaction object through the API interface to construct the financial data vector set Fin.

3. The financial data processing method based on big data analysis according to claim 2, characterized in that: S12. Account transaction data Dtr includes the total number of target transaction customers C, customer identifier c, number of transactions of the c-th customer nc, total number of transactions N, and standard deviation of the single transaction amount σAt; The business operation data Dre includes the total number of transaction channels M of the target transaction object, the channel identifier m, the time span Δtm from the establishment of the mth channel to the end of the observation period, the actual total revenue rm of the mth channel during the observation period, the expected revenue E[rm] of the mth channel calculated based on the historical database HIS, and the life cycle decay constant km of the mth channel; The production operation data Dop includes the average return rate Tp of the target transaction object within the preset statistical period T, the standard deviation of the return rate δTp within the preset statistical period T, the total number of sales units Nops, the number of production equipment Z, the equipment identifier z and the average annual failure rate λz of the zth equipment.

4. The financial data processing method based on big data analysis according to claim 3, characterized in that: S2 includes S21; S21. Based on the account transaction data Dtr in the financial feature vector set Fin, a transaction volatility assessment algorithm is constructed. A customer breadth correction factor is constructed based on the total number of customers C. Based on the customer identifier c and the total number of transactions N, and incorporating the Herfindahl-Hirschman index concept, the square of the ratio of the number of transactions nc of the cth customer to the total number of transactions N is calculated and the square root is taken. Finally, the standard deviation of the single transaction amount σAt is used as the transaction amount volatility factor to calculate the transaction volatility index Vt. The trading volatility evaluation algorithm expression is as follows: ； In the formula, ∑ represents the summation operation, Represents the square root operation.

5. The financial data processing method based on big data analysis according to claim 4, characterized in that: S3 includes S31; S31. Based on the business operation data Dre in the financial feature vector set Fin, construct a revenue channel authenticity assessment algorithm. Calculate the ratio of the actual total revenue rm of the mth channel during the observation period to the expected revenue E[rm] of the mth channel, perform an absolute value operation on the logarithm, and calculate the negative product of the time span Δtm from the establishment of the mth channel to the end of the observation period and the life cycle decay constant km of the mth channel. Input this into the natural exponential function exp. Combined with the total number of transaction channels M, calculate the revenue channel authenticity score Ra. The income channel authenticity evaluation algorithm expression is as follows: ； In the formula, |·| represents the absolute value operation, ln represents the logarithm operation with the natural constant e as the base, exp represents the natural exponential function, and ε represents the regularization term to prevent the denominator from being zero, specifically 1*10 -8 .

6. The financial data processing method based on big data analysis according to claim 5, characterized in that: S4 includes S41; S41. Based on the production and operation data Dop in the financial feature vector set Fin, a potential drag value assessment algorithm is constructed. A normal distribution model is constructed using the average return rate Tp and the standard deviation of the return rate δTp within a preset statistical period T as parameters, and the result is multiplied by the single return cost src and the total number of units sold Nops. A Poisson process model is applied to the annual average failure rate λz of the zth device and the preset statistical period T, and the result is multiplied by the preset economic loss Fz for a single failure of the zth device. The potential drag value Ld is calculated by combining the normal distribution model terms with the Poisson process model terms. Among them, the potential drag value evaluation algorithm expression is as follows: ； In the formula, src represents the average cost of a single return preset by cost accounting experts, p0 represents the preset benchmark return rate, and x represents the integral independent variable, which specifically means the future return rate in the interval [p0, 1]. represents the square root operation, Indicates the use of integral operation to predict the total probability that the integral independent variable x falls within the interval [p0, 1]. π represents the ratio of pi, which is a double-precision floating-point number with 16 significant digits. Fz represents the preset economic loss of a single failure of the zth device obtained from analysis of the historical database HIS. exp represents the natural exponential function.

7. The financial data processing method based on big data analysis according to claim 6, characterized in that: S5 includes S51 and S52; S51. Construct a financial feature vector set Fan based on the transaction volatility index Vt, the revenue channel authenticity score Ra, and the potential drag value Ld, and construct a comprehensive risk assessment algorithm. Standardize the transaction volatility index Vt and the revenue channel authenticity score Ra using the transaction volatility index reference value Vref and the revenue channel authenticity score reference value Rref. Compare the potential drag value Ld with the one-year EBITDA G of the transaction target. After eliminating dimensional differences, perform a weighted addition, and input the weighted summation result into the hyperbolic tangent function tanh to calculate the comprehensive risk index IRI. Among them, by analyzing the transaction volatility index Vt_HIS and the income channel authenticity score Ra_HIS in the historical database HIS, the transaction volatility index reference value Vref and the income channel authenticity score reference value Rref are obtained. The comprehensive risk assessment algorithm expression is as follows: ； Wherein, α represents the influencing factor of the transaction volatility index Vt obtained by performing multiple regression analysis on the historical database HIS, β represents the influencing factor of the income channel authenticity score Ra obtained by performing multiple regression analysis on the historical database HIS, and γ represents the influencing factor of the potential drag value Ld obtained by performing multiple regression analysis on the historical database HIS.

8. The financial data processing method based on big data analysis according to claim 7, characterized in that: S52. Compare the comprehensive risk index IRI with the decision threshold interval θ, generate a transaction decision AIR based on the comparison result, and execute the transaction decision; Among them, the decision threshold interval θ includes the first-level decision threshold θ1 and the second-level decision threshold θ2; If the comprehensive risk index IRI is less than the first-level decision threshold θ1, the target transaction object is determined to meet the transaction requirements, and the transaction decision AIR is to trade with the target transaction object in accordance with the agreed terms; If the first-level decision threshold θ1 ≤ comprehensive risk index IRI < second-level decision threshold θ2, the target transaction object is judged to have transaction risks but meet the transaction requirements. The transaction decision AIR is to renegotiate with the transaction target, request a reduction in the acquisition valuation and include a seller guarantee clause in the acquisition agreement. If the first-level decision threshold θ1 ≥ the second-level decision threshold θ2, it is determined that the target transaction object has transaction risks and does not meet the transaction requirements, and the transaction decision AIR is to give up the transaction with the target transaction object.

9. The financial data processing method based on big data analysis according to claim 8, characterized in that: S6 includes S61; S61. After executing the transaction decision AIR, continuously collect the financial data vector set Fin and actual profit amount APA of the target transaction object within one year; If the actual profit APA of the target transaction object in the transaction within one year is greater than the preset expected profit EPA, or the actual profit APA of the target transaction object in the transaction that is abandoned within one year is less than the preset expected profit EPA, then the transaction decision is recorded as correct; If the actual profit APA of the target transaction object within one year is less than the preset expected profit EPA, or the actual profit APA of the target transaction object within one year is greater than the preset expected profit EPA, then it will be recorded as a transaction decision error false; Calculate the trading decision accuracy rate AIR_true based on the number of correct trading decisions (true) and the number of incorrect trading decisions (false). Analyze the trading decision accuracy rate AIR_true every quarter. If the transaction decision accuracy rate AIR_true ≥ the preset expected transaction decision accuracy rate E_true, then there is no need for iterative optimization; If the transaction decision accuracy rate AIR_true is less than the preset expected transaction decision accuracy rate E_true, iterative optimization is required; Professionals in this field adjust the influencing factor α of the transaction volatility index Vt, the influencing factor β of the income channel authenticity score Ra, and the influencing factor γ of the potential drag value Ld in the comprehensive risk assessment algorithm, and conduct a multivariate regression analysis based on the historical database HIS and the transaction decision accuracy rate AIR_true. The adjusted influencing factor α of the transaction volatility index Vt, the influencing factor β of the income channel authenticity score Ra, and the influencing factor γ of the potential drag value Ld are calibrated again, and the financial feature vector set Fin, the financial feature vector set Fan, and the comprehensive risk index IRI are stored in the historical database HIS.

10. A financial data processing system based on big data analysis, applied to the financial data processing method based on big data analysis according to any one of claims 1 to 9, characterized in that: It includes financial data collection module, transaction fluctuation analysis module, revenue channel authenticity analysis module, potential drag value analysis module, comprehensive risk index analysis and evaluation module, and iterative optimization module; The data collection module collects accounting transaction data Dtr, business operation data Dre, and production operation data Dop from the target transaction object through the API interface to construct the financial data vector set Fin; The transaction volatility analysis module constructs a transaction volatility assessment algorithm based on the account transaction data Dtr in the financial feature vector set Fin, calculates the transaction volatility index Vt, and evaluates the stability and health of income; The revenue channel authenticity analysis module constructs a revenue channel authenticity evaluation algorithm based on the business operation data Dre in the financial feature vector set Fin, calculates the revenue channel authenticity score Ra, and evaluates the stability and health of the revenue growth model; The potential drag value analysis module constructs a potential drag value assessment algorithm based on the production operation data Dop in the financial feature vector set Fin to calculate the potential drag value Ld caused by equipment failure and product returns; The comprehensive risk index analysis and assessment module constructs a financial feature vector set Fan based on the transaction volatility index Vt, the revenue channel authenticity score Ra, and the potential drag value Ld. It then calculates the comprehensive risk index IRI and compares it with the decision threshold interval θ. Based on the comparison results, it generates and executes the transaction decision AIR. After the iterative optimization module completes the transaction decision AIR, it continuously collects the financial data vector set Fin and actual profit amount APA of the target transaction object and analyzes the transaction decision accuracy AIR_true. It performs iterative optimization based on the transaction decision accuracy AIR_true and stores the financial feature vector set Fin, financial feature vector set Fan and comprehensive risk index IRI in the historical database HIS.

Images