Enterprise credit investigation method based on innovative point system

Abstract

The invention relates to the technical field of enterprise credit investigation, in particular to an enterprise credit investigation method based on an innovation point system, and the method specifically comprises the following steps: 1, setting first-level indexes and weights of innovation points, the first-level indexes comprising innovation investment, innovation achievements, innovation management and market influence; 2, refining a first-level index of the innovation integral to obtain a second-level index; 3, calculating the score of each secondary index; 4, determining the weight of each secondary index; 5, calculating the score of each first-level index; 6, calculating a comprehensive score of the innovation points; step 7, carrying out credit rating division; 8, setting a dynamic adjustment mechanism; step 9, combining the macroeconomic factors into credit investigation; and 10, carrying out traditional credit data integration. According to the invention, by introducing an innovative point system, a more comprehensive enterprise credit investigation system is constructed, and fair credit evaluation is provided for innovative enterprises.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of enterprise credit investigation, and particularly to an enterprise credit investigation method based on an innovation credit system. BACKGROUND

[0002] Enterprise credit investigation is crucial for stakeholders such as financial institutions, partners, and government regulators. Traditional enterprise credit investigation methods mainly rely on financial data of enterprises, such as balance sheets, income statements, and cash flow statements, to assess the credit status of enterprises by analyzing these data. For example, a common credit scoring model such as the Z-score model has the formula: Z = 1.2X1 + 1.4X2 + 3.3X3 + 0.6X4 + 0.999X5. Wherein, X1 = working capital / total assets, reflecting the short-term solvency of the enterprise; X2 = retained earnings / total assets, reflecting the cumulative profitability of the enterprise; X3 = pre-tax profit / total assets, measuring the asset profitability of the enterprise; X4 = market value of shareholder equity / total liabilities, reflecting the financial structure of the enterprise; X5 = sales revenue / total assets, showing the asset operating capacity of the enterprise. This model based on financial data can reflect the credit risk of enterprises to some extent, but has obvious limitations.

[0003] Innovation is becoming increasingly critical to the survival and development of enterprises. Many innovative enterprises, especially those in the early stages or growth stages, have great innovation potential and development prospects, but due to the large amount of funds invested in R&D in the early stage, the financial data may not be ideal, resulting in a low credit rating under the traditional credit investigation system, making it difficult to obtain the support of financial institutions and the trust of partners. This not only limits the development of innovative enterprises, but also is not conducive to the innovation vitality and sustainable development of the entire economy. SUMMARY

[0004] In view of the technical problem in the prior art that innovative enterprises, especially those in the early stages or growth stages, due to the large amount of funds invested in R&D in the early stage, the financial data may not be ideal, resulting in a low credit rating under the traditional credit investigation system, the present application provides an enterprise credit investigation method based on an innovation credit system.

[0005] The technical solution adopted by the present application is: an enterprise credit investigation method based on an innovation credit system, which specifically comprises the following steps:

[0006] Step one: set up a first-level index and weight of innovation credit, the first-level index including innovation input, innovation achievement, innovation management, and market influence;

[0007] Step two: refine the first-level index of innovation credit to obtain a second-level index;

[0008] Step three: calculate the score of each second-level index;

[0009] Step four: determine the weight of each secondary indicator;

[0010] Step five: calculate the score of each primary indicator;

[0011] Step six: calculate the comprehensive score of innovation points;

[0012] Step seven: divide the credit rating;

[0013] Step eight: set up a dynamic adjustment mechanism;

[0014] Step nine: combine macroeconomic factors into credit investigation;

[0015] Step ten: integrate traditional credit investigation data;

[0016] Step eleven: build a credit risk early warning model;

[0017] Step twelve: establish an enterprise innovation competitiveness evaluation model.

[0018] In one embodiment, in step one: the specific method of setting the primary indicators and weights of innovation points is as follows:

[0019] Let the set of primary indicators be I = {I1, I2, I3, I4}, I1 represents innovation input, I2 represents innovation achievement, I3 represents innovation management, and I4 represents market influence. The corresponding weight vector is W 1 , and the calculation is as follows:

[0020]

[0021] Among them,

[0022]

[0023] The method of determining the weight adopts the analytic hierarchy process, which is as follows:

[0024] First, construct the judgment matrix A. For the four factors of innovation input I1, innovation achievement I2, innovation management I3 and market influence I4, compare their relative importance two by two. Let the judgment matrix be:

[0025]

[0026] Among them, a ij represents the importance scale of factor i relative to factor j, and the value range is 1-9 and its inverse, 1 means equal importance, 3 means the former is slightly more important than the latter, 5 means the former is obviously more important than the latter, 7 means the former is strongly more important than the latter, and 9 means the former is extremely more important than the latter. 2, 4, 6, 8 are intermediate values, among which,

[0027] The maximum eigenvalue λ of the judgment matrix A is calculated max and its corresponding eigenvector W, and consistency check is performed;

[0028] The consistency index CI is calculated as follows:

[0029]

[0030] Where n is the order of the matrix, and the random consistency index RI is obtained from the standard table according to the order of the matrix;

[0031] The consistency ratio CR When CR < 0.1, it is considered that the judgment matrix has satisfactory consistency, and the eigenvector W obtained at this time is the weight vector W of each factor 1 .

[0032] In one of the embodiments, in step two: refining the first-level indicators of innovation points to obtain second-level indicators, including specific refinement of innovation input, innovation achievement, innovation management, and market influence, and the specific calculation method is as follows:

[0033] Among them, the innovation input includes R&D fund input and R&D personnel input;

[0034] Let the R&D fund input of the enterprise in the tth year be R t , and the proportion of the current year's operating income be Among them, O t is the operating income of the enterprise in the tth year;

[0035] Let the number of R&D personnel of the enterprise be P, and the proportion of the total number of employees T be

[0036] Among them, the innovation achievement includes the number of patent applications and the sales revenue of new products;

[0037] Let the number of patents applied for by the enterprise in the tth year be N a,t , the number of patents granted be N g,t , and the patent conversion rate be r p,t ;

[0038]

[0039] Among them, when N a,t = 0, r p,t = 0;

[0040] The sales revenue of new products of the enterprise in the tth year is S n,t , and the proportion of the total sales revenue in the current year is r S,n,t , which is calculated as follows:

[0041]

[0042] wherein S t is the total sales revenue of the enterprise in the tth year;

[0043] wherein the innovation management includes the innovation strategy planning perfection degree and the effectiveness of the innovation incentive system;

[0044] The innovation strategy planning perfection degree is scored, and the score of the innovation strategy planning perfection degree of the enterprise is denoted as S s ;

[0045] The effectiveness of the innovation incentive system is scored, and the score of the effectiveness of the innovation incentive system of the enterprise is denoted as S i ;

[0046] wherein the market influence includes the brand awareness and the market share;

[0047] The brand awareness B k of the enterprise in the target market is calculated through market research;

[0048] The market share M s of the enterprise in the industry is obtained through inquiry, and the following calculation is performed:

[0049]

[0050] wherein Q is the sales volume of the product or service of the enterprise, and Q total is the total sales volume of the industry.

[0051] In one embodiment, the calculation of the secondary index score in step three includes the R&D fund input score, the R&D personnel input score, the patent application quantity score, the patent conversion rate score, the new product sales revenue score, the innovation strategy planning perfection degree score, the market share score, the brand awareness score, and the effectiveness of the innovation incentive system score;

[0052] The calculation method of the R&D fund input score is as follows:

[0053] Let the R&D fund input score function be f R (r R,t ), f R (r R,t ) = α1 × r R,t , wherein α1 is an adjustment coefficient;

[0054] The calculation method of the R&D personnel input score is as follows: Let the R&D personnel input score function be f P (r P ), f P (r P ) = α2 × r P , wherein α2 is an adjustment coefficient;

[0055] The calculation method of the patent application quantity score is as follows: assuming that the patent application quantity score function is f Wherein β1 is an adjustment coefficient;

[0056] The calculation method of the patent conversion rate score is as follows: assuming that the patent conversion rate score function is f p (r p,t ), f p (r p,t ) = β2 × r p,t , wherein β2 is an adjustment coefficient;

[0057] The calculation method of the new product sales revenue score is as follows: assuming that the new product sales revenue score function is f Wherein γ1 is an adjustment coefficient;

[0058] The calculation method of the innovation strategy planning perfection degree score is as follows: a scoring method is adopted, and the scored score S s is taken as the score;

[0059] The calculation method of the innovation incentive system effectiveness score is as follows: a scoring method is adopted, and the scored score S i is taken as the score;

[0060] The calculation method of the brand awareness score is as follows: assuming that the brand awareness score function is f B (B k ), f B (B k ) = δ1 × B k , wherein δ1 is an adjustment coefficient;

[0061] The calculation method of the market share score is as follows: assuming that the market share score function is f M (M s ), f M (M s ) = δ2 × M s , wherein δ2 is an adjustment coefficient;

[0062] Wherein, in step four, the specific method for determining the weight of the secondary indicators is as follows:

[0063] For each secondary indicator under each primary indicator, the weight is determined by using the analytic hierarchy process;

[0064] Assuming that the weight vectors corresponding to the secondary indicators of innovation input, i.e. R&D fund input and R&D personnel input, are

[0065] and

[0066] Build judgment matrix and calculate weight, the same reason, determine the weight vector of the secondary indicators under the innovation achievements, innovation management and market influence and

[0067] In one embodiment, in step five: the specific method of calculating the score of the primary indicators is as follows:

[0068] The score of innovation input is calculated as follows:

[0069] The score of innovation achievements is calculated as follows:

[0070] The score of innovation management is calculated as follows: 4. Market influence score:

[0071] The score of market influence is calculated as follows:

[0072] In step six: the specific method of calculating the comprehensive score of innovation points is as follows:

[0073] Let the comprehensive score of innovation points be S total , then

[0074] In step seven: the specific method of dividing credit grades is as follows:

[0075] According to the comprehensive score of innovation points S total , divide the credit grades;

[0076] Let the set of credit grades be C={C1, C2, C3, C4, C5}, corresponding to different credit grades respectively;

[0077] In step eight: the specific method of setting up dynamic adjustment mechanism is as follows:

[0078] Let the adjustment period be T;

[0079] At the end of each adjustment period, collect the relevant data of enterprises again, and calculate the comprehensive score of innovation points again

[0080] Let the adjustment coefficient be θ, then the adjusted comprehensive score of innovation points is The value range of θ is (0, 1).

[0081] In one embodiment, step nine: the specific method of combining macroeconomic factors into credit investigation is as follows:

[0082] Select macroeconomic indicators closely related to business operations;

[0083] Construct a vector autoregressive model to analyze the dynamic relationship between macroeconomic indicators and enterprise credit indicators. The specific method is as follows:

[0084] Let the enterprise credit score CS, GDP growth rate GDP g , inflation rate IR, and interest rate level R form a vector Y t = (CS t , GDP g,t , IR t , R t ) T ;

[0085] The expression of the vector autoregressive model is where Φ i is the coefficient matrix, p is the lag order, and ∈ t is the random error term;

[0086] By estimating the parameters of the vector autoregressive model, the influence coefficients of macroeconomic indicators on enterprise credit scores are obtained;

[0087] In step ten, the specific method for integrating traditional credit data is as follows:

[0088] First, filter traditional credit data, including bank credit repayment records and commercial credit transaction records;

[0089] Calculate the asset-liability ratio AR, current ratio CR, net asset return rate ROE, Total liabilities is the total liabilities of the enterprise, Total assets is the total assets of the enterprise, Current assets is the current assets of the enterprise, Current liabilities is the current liabilities of the enterprise, Net incom e is the net profit of the enterprise, and Average equity is the average shareholder equity of the enterprise;

[0090] Next, use factor analysis to reduce the dimensionality of traditional credit data. Let the traditional credit data matrix be Y = (y ij ) n×p , where n is the number of enterprise samples and p is the number of traditional credit indicators;

[0091] After standardizing Y, calculate its correlation coefficient matrix R;

[0092] The eigenvalues ​​of the correlation coefficient matrix R are obtained through eigenvalue decomposition: λ1≥λ2≥…≥λ p and the corresponding eigenvectors u1, u2, ..., u p ;

[0093] Select the first q eigenvectors to construct the factor loading matrix A = (u1, u2, ..., u q );

[0094] The dimensionality-reduced traditional credit data Z = YA is then concatenated with the innovation score S to form a comprehensive credit assessment data matrix M = (Z|S).

[0095] A credit scoring model is constructed using the random forest algorithm. The comprehensive credit assessment data matrix M is divided into a training set and a test set. On the training set, the parameters of the random forest are adjusted, and a grid search combined with cross-validation is used to find a suitable parameter combination.

[0096] During cross-validation, the average accuracy, recall, and F1 score of the model are calculated as evaluation metrics under different parameter combinations.

[0097] The trained randomized Rylin model was evaluated using a test set.

[0098] The accuracy of the calculation model. Recall in, The evaluation assesses the model's ability to classify enterprises with different credit ratings. TP represents true positives, indicating the number of samples correctly predicted as positive; TN represents true negatives, indicating the number of samples correctly predicted as negative; FP represents false positives, indicating the number of samples incorrectly predicted as positive; and FN represents false negatives, indicating the number of samples incorrectly predicted as negative.

[0099] In one embodiment, step eleven involves the following specific method for constructing a credit risk early warning model:

[0100] The debt-to-equity ratio (AR) and current ratio (CR) are standardized, and the original values ​​are set as x. i The standardized value is Using formula in Let σ be the mean of the indicator, and σ be the standard deviation of the indicator.

[0101] A credit risk early warning model is constructed using a logistic regression model. Let the early warning result be Y (where 0 represents low risk and 1 represents high risk), and the influencing factors be X1, X2, ..., X... n(Including innovation points-related indicators and traditional financial indicators), the logistic regression model formula is as follows:

[0102]

[0103] Where β0, β1, ..., β n The model parameters are estimated using the maximum likelihood estimation method;

[0104] Early warning threshold determination: Through analysis of a large amount of historical data, a suitable early warning threshold P0 is determined; when P(Y=1|X1,X2,…,X…) n When )≥P0, a credit risk warning signal is issued.

[0105] In one embodiment, step twelve involves the following specific method for establishing an enterprise innovation competitiveness assessment model:

[0106] Based on the innovation score, we further analyze the company's innovation speed, the uniqueness of its innovation, and the sustainability of its innovation;

[0107] The speed of innovation is measured by the frequency, F, of launching new products.

[0108] The uniqueness of an innovation is reflected in the novelty assessment of a patent;

[0109] The sustainability of innovation is reflected by the stability of R&D investment, specifically using the coefficient of variation (CV) of R&D investment. R , Where σ R The standard deviation of R&D investment (The average R&D investment);

[0110] These new indicators will be combined with the various indicators in the innovation score to construct a comprehensive indicator system for evaluating corporate innovation competitiveness;

[0111] The relative weights between the newly included indicators and the original innovation score indicators were determined using the analytic hierarchy process.

[0112] A judgment matrix B is constructed based on seven factors: innovation speed, uniqueness, sustainability, innovation investment, innovation results, innovation management, and market influence.

[0113]

[0114] Among them, b ij The meaning of a ij Similarly, it represents the importance scale of factor i relative to factor j;

[0115] By calculating the largest eigenvalue of the matrix and its corresponding eigenvector, and performing a consistency check, the weight vectors of each factor are obtained, as follows:

[0116] and

[0117] Calculate the innovation competitiveness score: Let the enterprise's innovation competitiveness score be S. comp The calculation formula is:

[0118]

[0119] The beneficial effects of this invention are as follows: Compared with the prior art, this invention introduces an innovation points system, which comprehensively considers multiple dimensions of factors such as enterprise innovation input, innovation results, and innovation management, to build a more comprehensive enterprise credit system, providing fair credit evaluation for innovative enterprises, promoting the rational allocation of financial resources to the innovation field, and driving enterprise innovation development and economic structural transformation and upgrading. Attached Figure Description

[0120] Figure 1 This is a flowchart illustrating the present invention. Detailed Implementation

[0121] In the description of this invention, it should be noted that the terms "front", "up", "down", "left", "right", "vertical", "horizontal", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0122] To address the problems existing in the background technology, this application proposes the following technical solution: a corporate credit investigation method based on an innovation points system, specifically including the following steps:

[0123] Step 1: Establish the primary indicators and weights for the innovation points. The primary indicators include innovation input, innovation results, innovation management and market influence. There is no limit to the number of primary indicators. In this example, four are set as a reference.

[0124] Step 2: Refine the primary indicators of the innovation points to obtain secondary indicators;

[0125] Step 3: Calculate the scores for each secondary indicator;

[0126] Step 4: Determine the weight of each secondary indicator;

[0127] Step 5: Calculate the scores for each primary indicator;

[0128] Step Six: Calculate the overall innovation score;

[0129] Step 7: Determine credit rating;

[0130] Step 8: Establish a dynamic adjustment mechanism;

[0131] Step Nine: Incorporate macroeconomic factors into credit reporting;

[0132] Step 10: Integrate traditional credit data;

[0133] Step 11: Construct a credit risk early warning model;

[0134] Step 12: Establish an enterprise innovation competitiveness assessment model.

[0135] The above technical solutions are explained in turn as follows:

[0136] In Step One: The specific method for establishing the primary indicators and weights for the innovation score is as follows:

[0137] Let the set of primary indicators be I = {I1, I2, I3, I4}, where I1 represents innovation input, I2 represents innovation output, I3 represents innovation management, and I4 represents market influence. The corresponding weight vector is W. 1 And the calculation is as follows:

[0138]

[0139] in,

[0140]

[0141] The weights are determined using the analytic hierarchy process (AHP), as detailed below:

[0142] First, construct a judgment matrix A. For the four factors—innovation input I1, innovation results I2, innovation management I3, and market influence I4—compare their relative importance pairwise. Let the judgment matrix be:

[0143]

[0144] Among them, a ij The importance scale represents factor i relative to factor j, with values ​​ranging from 1 to 9 and their reciprocals. 1 indicates that both are equally important, 3 indicates that factor i is slightly more important than factor j, 5 indicates that factor i is significantly more important than factor j, 7 indicates that factor i is strongly more important than factor j, and 9 indicates that factor i is extremely more important than factor j. Values ​​2, 4, 6, and 8 are intermediate values.

[0145] In a further design, the largest eigenvalue λ of the judgment matrix A is calculated. maxAnd its corresponding eigenvector W, and perform a consistency check;

[0146] The consistency index (CI) is calculated as follows:

[0147]

[0148] Where n is the matrix order, and the random consistency index RI is obtained from the standard table based on the matrix order;

[0149] Consistency ratio When CR < 0.1, the judgment matrix is ​​considered to have satisfactory consistency, and the resulting eigenvector W is the weight vector W of each factor. 1 .

[0150] The above technical solution can be explained as follows: For example, when evaluating a technology company, suppose that innovation input I1 is slightly more important than innovation output I2, so a 12 =3, then a 21 =1 / 3; Innovation input I1 is significantly more important than innovation management I3, a 13 =5, then a 31 =1 / 5; Innovation investment I1 is significantly more important than market influence I4, a 14 =7, then a 41 =1 / 7; Innovation Achievements I2 is slightly more important than Innovation Management I3, a 23 =3, then a 32 =1 / 3; Innovation achievements I2 are significantly more important than market influence I4, a 24 =5, then a 42 =1 / 5; Innovation Management I3 is slightly more important than Market Influence I4, a 34 =3, then a 43 =1 / 3, resulting in the judgment matrix A:

[0151]

[0152] Then, the largest eigenvalue of the judgment matrix A and its corresponding eigenvector are calculated, and a consistency check is performed (assuming that the calculation meets the consistency requirements).

[0153] Next, scores for innovation input, innovation achievements, innovation management, and market influence are calculated based on the actual situation of the enterprise, and then the overall innovation score is calculated by combining the weights.

[0154] For example, if a company scores 80 points for innovation input, 70 points for innovation achievements, 60 points for innovation management, and 50 points for market influence, its overall innovation score is 72.5 points.

[0155] In this embodiment, because the evaluation considers multiple dimensions such as innovation investment and innovation results, and the weights are reasonably determined, it avoids the one-sidedness of focusing only on a single dimension such as financial data, and can more comprehensively reflect the true credit status of enterprises. For example, some enterprises in the early stages of development but with large innovation investment and high potential for innovation results can receive a fairer evaluation under this assessment system. This is conducive to enterprises gaining support from financial institutions and the trust of partners, promoting the innovative development of enterprises, and also conducive to the rational allocation of resources to enterprises with greater innovation potential, thus driving innovation and progress in the entire industry.

[0156] In step two: the primary indicators of the innovation score are refined to obtain secondary indicators, including specific details on innovation input, innovation results, innovation management, and market influence, and the specific calculation method is as follows:

[0157] Innovation investment includes R&D funding and R&D personnel investment;

[0158] Let R be the company's R&D investment in year t. t The proportion of the annual operating revenue was Among them O t Let t be the company's operating revenue in year t.

[0159] Let P be the number of R&D personnel in the company, and let T be the proportion of the company's total employees.

[0160] These innovative achievements include the number of patent applications and sales revenue from new products;

[0161] Let N be the number of patents filed by the company in year t. a,t The number of patents granted is N g,t The patent conversion rate is r p,t ;

[0162]

[0163] Where, when N a,t When r = 0, p,t =0;

[0164] The company's sales revenue from its new products in year t is S. n,t The proportion of total sales revenue in that year was r S,n,t The calculation is as follows:

[0165] Among them, S t Let t be the company's total sales revenue in year t.

[0166] The above technical solution is explained as follows: By refining the primary indicators of the innovation score, secondary indicators are derived to more accurately measure a company's innovation performance. Regarding innovation investment, quantification is applied across two dimensions: R&D funding and R&D personnel. This comprehensively reflects the company's support for innovation in terms of human resources and capital, allowing companies to clearly understand their strengths and weaknesses in innovation resource investment and facilitating comparisons between different companies. In terms of innovation achievements, the number of patent applications and conversion rate reflect the quantity and quality of the company's innovation output, while the proportion of new product sales revenue reflects the degree to which the market value of innovation achievements is realized. This helps companies assess the actual effectiveness of their innovation activities and provides external investors and partners with a direct basis for understanding the commercial value of the company's innovation achievements. Overall, this refined solution, through multi-dimensional and quantitative indicator settings, more comprehensively, deeply, and accurately assesses a company's innovation capabilities and innovation value, overcoming the potential biases and ambiguities of previous assessments. It provides more valuable references for companies' self-improvement, external cooperation, and resource allocation, helping companies develop better in an innovation-driven market environment.

[0167] Innovation management includes the completeness of innovation strategic planning and the effectiveness of innovation incentive systems;

[0168] The scoring method can employ either human experts or machine scoring, with a maximum score of ten. The score for the completeness of a company's innovation strategy plan is denoted as S. s ;

[0169] The effectiveness of an innovation incentive system is assessed using a scoring method, which can be achieved through human experts or machine scoring. The maximum score is ten, and the company's effective innovation incentive system receives an S score. i ;

[0170] The method for machine scoring is as follows:

[0171] First, we collect a large number of innovation strategy planning documents from companies of different industries and sizes. These documents are then preprocessed, including text cleaning (removing noise, special characters, etc.) and word segmentation.

[0172] Extract key features, such as:

[0173] Strategic Goal Clarity: Text analysis techniques are used to determine whether the plan clearly articulates long-term and short-term innovation goals, such as specific market share targets and technological breakthroughs. Named entity recognition technology from natural language processing can be used to identify target-related keywords and quantify their frequency and location.

[0174] Resource allocation rationality: Analyze the allocation of resources such as R&D funds and human resources in the plan. The rationality of resource allocation can be assessed by identifying resource-related words and phrases in the text, such as "R&D investment ratio" and "recruitment plan," combined with numerical calculations and semantic analysis.

[0175] Risk Response Strategies: Examine the plan for analyses and countermeasures regarding potential risks (such as technological and market risks). Sentiment analysis and topic modeling can be used to identify risk-related content and assess the completeness and feasibility of the response strategies.

[0176] Model selection and training:

[0177] Various machine learning models can be chosen, such as Support Vector Machines (SVM), Random Forests, and Neural Networks. Taking neural networks as an example, let's build a multi-layered text classification model.

[0178] The extracted features are used as input, and expert ratings for these documents are collected as labels to form a training dataset. The model is trained using the training dataset, and the model parameters are adjusted through the backpropagation algorithm to make the model's predictions as close as possible to the expert ratings.

[0179] Model Evaluation and Application:

[0180] The trained model is evaluated using a test dataset. Common evaluation metrics include mean squared error (MSE) and mean absolute error (MAE). If the model performs as expected, it can be applied to score new corporate innovation strategy planning documents. The features of the new document are input into the model, and the model's output is the machine score of the completeness of the corporate innovation strategy plan.

[0181] Data collection and feature extraction:

[0182] Collect relevant documents regarding the company's innovation incentive system, such as employee handbooks and incentive policy documents. Perform text preprocessing as well.

[0183] Extract key features, such as:

[0184] Incentive diversity: This analysis examines the incentive methods included in the system, such as bonuses, promotion opportunities, and equity. Text mining techniques are used to statistically analyze the frequency and proportion of different incentive methods to measure their diversity.

[0185] Clarity of Incentive Standards: This assesses whether the standards for employee incentives are clearly defined within the incentive system, such as performance indicators and requirements for innovative achievements. Text matching and semantic analysis techniques can be used to evaluate the clarity of these standards.

[0186] Employee Feedback and Results: Collect feedback data from internal employees regarding the incentive system, such as survey results and employee satisfaction ratings. Combine this data with the incentive system document to extract relevant features, such as the correlation between employee satisfaction and incentive system clauses.

[0187] Model selection and training:

[0188] Alternatively, you can choose a suitable machine learning model, such as a linear regression model (if you want to output a continuous score) or a classification model (such as classifying the effectiveness into high, medium, and low levels).

[0189] Taking a linear regression model as an example, the extracted features are used as independent variables, and the experts' scores for the effectiveness of the incentive system are used as the dependent variable to form a training dataset for training. The model parameters are adjusted using optimization algorithms such as the least squares method, enabling the model to accurately predict the effectiveness score of the incentive system.

[0190] Model Evaluation and Application:

[0191] The model is evaluated using an independent test dataset to ensure its generalization ability. After passing the evaluation, the relevant features of the new company's incentive system are input into the model, and the model output is the machine score of the effectiveness of the company's innovation incentive system.

[0192] By using the above methods, machine models can be used to score the completeness of innovation strategic planning and the effectiveness of innovation incentive systems in a more objective and accurate manner, thereby improving the efficiency and consistency of scoring and reducing the interference of human factors.

[0193] Market influence includes brand awareness and market share.

[0194] Through market research, calculate the brand awareness rate (B) of the enterprise in the target market. k ;

[0195] The company's market share in its industry is M. s The calculation is as follows:

[0196]

[0197] Where Q represents the sales volume of the company's products or services. total This represents the total sales volume of the industry.

[0198] The above technical solution is explained as follows: Market influence is assessed by refining it into two key indicators: brand awareness and market share. Market research calculating brand awareness in the target market directly reflects the brand's reach and recognition among the target audience, helping companies understand the effectiveness of their brand promotion. It also provides external investors and partners with a quantitative reference for brand influence, facilitating the assessment of the company's potential competitiveness in the market. Market share, measured as the ratio of a company's product or service sales to the total sales of its industry, clearly demonstrates the company's market position. This not only helps companies clarify their industry location and identify market expansion opportunities and challenges but also allows stakeholders to assess the company's market competitiveness and development potential, providing a strong basis for investment and cooperation decisions. Overall, this solution, through quantitative and targeted indicator settings, more comprehensively and accurately assesses a company's market influence, overcoming the ambiguity and subjectivity that may exist in previous market assessments.

[0199] Step 3: The calculation of secondary indicator scores includes R&D funding investment score, R&D personnel investment score, number of patent applications score, patent conversion rate score, new product sales revenue score, innovation strategy planning completeness score, market share score, brand awareness score, and innovation incentive system effectiveness score.

[0200] The calculation method for the R&D funding investment score is as follows:

[0201] Let the scoring function for R&D funding investment be f. R (r R,t ), f R (r R,t )=α1×r R,t , where α1 is an adjustment coefficient; it can be determined based on factors such as the industry's average R&D investment ratio.

[0202] The calculation method for R&D personnel input score is as follows: Let the R&D personnel input score function be f. P (r P ), f P (r P )=α2×r P , where α2 is the adjustment coefficient;

[0203] The calculation method for the patent application quantity score is as follows: Let the patent application quantity score function be... (The logarithmic function is used to avoid the score from increasing too quickly when there are too many patents), where β1 is the adjustment coefficient;

[0204] The patent conversion rate score is calculated as follows: Let the patent conversion rate score function be f. p (r p,t ), f p (r p,t )=β2×r p,t , where β2 is the adjustment coefficient;

[0205] The calculation method for the new product sales revenue score is as follows: Let the new product sales revenue score function be... Where γ1 is the adjustment coefficient;

[0206] The above technical solution is explained as follows:

[0207] The above technical solution quantifies the scores of secondary indicators by constructing different scoring functions. For the R&D funding investment scoring function f... R (r R,t )=α1×r R,t The scoring function f of R&D personnel input P (r P )=α2×r P New product sales revenue scoring function Using a linear function, the relationship between input / output ratios and scores is presented simply and intuitively, facilitating understanding and calculation. This allows companies to clearly understand their performance and corresponding scores in these areas, and also facilitates comparisons between different companies. (Patent application quantity scoring function) Using a logarithmic function avoids excessively rapid score increases when there are too many patents, allowing the score to more reasonably reflect the actual value of a company's patent applications and preventing a situation where quantity is simply pursued while quality is neglected. Patent conversion rate scoring function f p (r p,t )=β2×r p,t This is directly related to the conversion rate, highlighting the importance of turning patents into practical results.

[0208] α1 (Adjustment coefficient for R&D investment): The value can be taken with reference to the industry average R&D investment ratio. If the industry average R&D investment ratio is high, such as in high-tech industries, α1 can be appropriately larger, such as between 1 and 3, to encourage companies to increase R&D investment; if the industry average R&D investment ratio is low, α1 can be taken between 0.5 and 1 to make the score more in line with the actual situation of the industry.

[0209] α2 (Adjustment coefficient for R&D personnel input): The value can be determined based on the industry's dependence on R&D personnel. For industries that are highly dependent on R&D personnel, such as the software R&D industry, α2 can be between 1 and 2; for traditional industries with a relatively small proportion of R&D personnel, α2 can be between 0.3 and 0.8.

[0210] β1 (Patent Application Quantity Adjustment Factor): This can be determined based on the industry's innovation activity and patent value. In industries with high innovation activity and high patent value, such as the biopharmaceutical industry, β1 can be between 0.5 and 1; in industries with a relatively large number of patents but relatively low value, β1 can be between 0.2 and 0.5.

[0211] β2 (Patent Conversion Rate Adjustment Coefficient): The value can be determined based on the general difficulty and importance of patent conversion within the industry. If patent conversion is difficult and crucial to enterprise development, such as in high-end manufacturing, β2 can be between 1 and 2; if patent conversion is relatively easy, β2 can be between 0.5 and 1.

[0212] γ1 (New Product Sales Revenue Adjustment Coefficient): This can be determined based on the level of market competition in the industry and the market potential of the new product. In industries with intense competition and high market potential for new products, such as the consumer electronics industry, γ1 can be between 1 and 3; in industries with relatively stable markets and less impact from new products, γ1 can be between 0.3 and 1.

[0213] The calculation method for the perfection score of the innovation strategy plan is as follows: a scoring method is used, and the score S is... s As a score;

[0214] The calculation method for the effectiveness score of the innovation incentive system is as follows: a scoring method is used, and the score S is... i As a score;

[0215] The scoring method mentioned above can be either human or machine-based, and the machine scoring method is consistent with the scoring principle used for the completeness of innovation strategic planning.

[0216] The brand awareness score is calculated as follows: Let the brand awareness score function be f. B (B k ), f B (B k )=δ1×B k , where δ1 is the adjustment coefficient;

[0217] The market share score is calculated as follows: Let the market share score function be f. M (M s ), f M (M s )=δ2×M s , where δ2 is the adjustment coefficient.

[0218] In step four, the specific method for determining the weights of the secondary indicators is as follows:

[0219] For each secondary indicator under a primary indicator, the weights are determined using the analytic hierarchy process (AHP).

[0220] Let the weight vectors for the secondary indicators of innovation input, namely R&D funding input and R&D personnel input, be .

[0221] and

[0222] Similarly, construct the judgment matrix and calculate the weights, and determine the weight vectors of the secondary indicators under innovation achievements, innovation management, and market influence. and

[0223] The above technical solution is explained as follows:

[0224] From the perspective of the scoring function, the brand awareness scoring function f B (B k )=δ1×B k With market share scoring function f M (M s )=δ2×M s This method converts market influence-related indicators into scores using a simple linear relationship, making it intuitive, easy to understand, and easy to calculate. It clearly demonstrates the direct correlation between brand awareness and market share and the score, allowing companies to understand their performance in market influence and facilitating quick access to relevant information for external assessors.

[0225] δ1 (Brand Awareness Adjustment Coefficient): The value can be determined based on the level of market competition in the industry and the importance of the brand. In highly competitive industries with significant brand effects, such as the fashion consumer goods industry, consumers' brand awareness and loyalty have a greater impact on purchasing decisions. δ1 can be set between 1 and 3 to highlight the importance of brand awareness. In contrast, in some technology-oriented industries, factors such as product performance are relatively more important, and the impact of brand awareness is relatively smaller. δ1 can be set between 0.3 and 1.

[0226] δ2 (Market Share Adjustment Coefficient): The value can be determined based on the industry's market structure and development stage. In monopolistic competition or oligopolistic industries, market share has a significant impact on a company's competitiveness and profitability, and δ2 can be between 1 and 3. In emerging industries or highly competitive companies, market share changes relatively frequently and have a relatively small impact on the company, and δ2 can be between 0.5 and 1.

[0227] In determining the weights of secondary indicators, the analytic hierarchy process (AHP) can comprehensively consider the relative importance of each secondary indicator within its primary indicator. By constructing a judgment matrix and calculating the weights, the determination of weights can be made more scientific and reasonable. Taking innovation investment as an example, the weight vectors for R&D funding and R&D personnel investment are determined as follows: This reflects the relative importance of these two factors within the dimension of innovation investment. Different industries have varying degrees of reliance on R&D funding and personnel; the analytic hierarchy process (AHP) can be used to allocate weights reasonably based on actual circumstances. Similarly, determining the weights of the secondary indicators under innovation achievements, innovation management, and market influence can make the overall innovation score calculation more closely reflect the actual situation of enterprises.

[0228] In step five, the specific method for calculating the score of the primary indicator is as follows:

[0229] The innovation input score is calculated as follows:

[0230] The score for innovative achievements is calculated as follows:

[0231] Innovation management score is calculated as follows: 4. Market Influence Score:

[0232] Market influence score is calculated as follows:

[0233] The above technical solution can be explained as follows: This solution comprehensively and systematically evaluates a company's innovation capabilities and overall performance. Through this hierarchical calculation system, it provides in-depth understanding of a company's specific situation in each innovation dimension, and also allows for a holistic grasp of the company's innovation level, providing clear directions for improvement. For example, if the innovation achievement score is low, the company can specifically strengthen its patent application and commercialization efforts, as well as new product promotion. It also provides external stakeholders, such as investors, partners, and government departments, with an objective and scientific evaluation basis, facilitating their informed decision-making. For instance, investors can use the comprehensive innovation score to select companies with innovation potential and promising development prospects for investment.

[0234] In step six, the specific method for calculating the comprehensive innovation score is as follows:

[0235] Let the overall innovation score be S. total ,but

[0236] In step seven, the specific method for credit rating is as follows:

[0237] Based on the comprehensive score of innovation points S total Credit rating;

[0238] Let the set of credit ratings be C = {C1, C2, C3, C4, C5}, each corresponding to a different credit rating. For example, when S... totalWhen the score is ≥80, the credit rating is C1 (Excellent); when the score is ≤65, the credit rating is C1 (Excellent). toal When the value is less than 80, the credit rating is C2 (good); when the value is less than or equal to 50, the credit rating is C2 (good). total When the value is less than 65, the credit rating is C3 (medium); when 35 ≤ S total When the value is less than 50, the credit rating is C4 (average); when the value is less than 50, the credit rating is C4 (average). total When the credit score is less than 35, the credit rating is C5 (poor).

[0239] In step eight, the specific method for establishing a dynamic adjustment mechanism is as follows:

[0240] A company's innovation capabilities and creditworthiness are dynamic and therefore require a dynamic adjustment mechanism.

[0241] Let the adjustment period be T (e.g., quarterly, semi-annually, or annually);

[0242] At the end of each adjustment period, relevant data from enterprises are collected again, and the overall innovation score is recalculated.

[0243] Let the adjustment coefficient be θ, then the adjusted innovation score is... The value of θ ranges from (0,1) and can be determined based on factors such as industry characteristics and the rate of market change. Credit ratings are then reclassified based on the adjusted scores.

[0244] The above technical solution is explained as follows: Clearly defining the adjustment cycle ensures that a company's innovation score and credit status are regularly updated and assessed, adapting to the dynamic changes in a company's innovation capabilities and the market environment. At the end of each adjustment cycle, data is re-collected and a new comprehensive innovation score is calculated, ensuring that the assessment results promptly reflect the company's latest innovation and credit status. The adjusted comprehensive innovation score is obtained by weighting the score before adjustment and the new score using an adjustment coefficient. This method considers both the company's past performance and current changes. For example, for innovative companies in a rapid development phase, if recent innovation achievements are significant, the θ value can be appropriately increased to give the new score a larger proportion of the comprehensive score, thereby promptly improving their credit rating; while for companies with relatively stable development, the θ value can be appropriately decreased to make the score adjustment more stable.

[0245] Step Nine: The specific methods for incorporating macroeconomic factors into credit reporting are as follows:

[0246] Select macroeconomic indicators closely related to business operations, such as GDP growth rate, inflation rate, and interest rate level.

[0247] Let the GDP growth rate be GDP gThe inflation rate is IR, and the interest rate is R. These indicators reflect the overall state of the macroeconomic environment and influence aspects such as market demand, cost structure, and financing environment for businesses.

[0248] A vector autoregression model is constructed to analyze the dynamic relationship between macroeconomic indicators and corporate credit indicators. The specific method is as follows:

[0249] Let the corporate credit score (CS) and GDP growth rate (GDP) be used as the basis for the calculation. g The vector Y consists of the inflation rate IR and the interest rate level R. t =(CS t GDP g,t IR t ,R t ) T ;

[0250] The expression for the vector autoregressive model is: Where Φ i Let p be the coefficient matrix, and p be the lag order, ∈ t This is the random error term;

[0251] By estimating the parameters of the vector autoregressive model, the impact coefficient of macroeconomic indicators on corporate credit scores can be obtained. For example, if the estimated impact coefficient of GDP growth rate on corporate credit scores is... When GDP growth rate increases, the change in corporate credit score is: Incorporating macroeconomic factors into credit assessments makes credit results more reflective of the impact of the macroeconomic environment on corporate credit.

[0252] The above technical solution is explained as follows: By selecting macroeconomic indicators closely related to business operations, such as GDP growth rate, inflation rate, and interest rate level, and constructing a vector autoregression model, the dynamic relationship between macroeconomic indicators and corporate credit indicators is analyzed. First, incorporating macroeconomic factors into the credit assessment system allows credit results to better reflect the true impact of the macroeconomic environment on corporate credit. Changes in the macroeconomic environment, such as GDP growth, inflation, and interest rate fluctuations, directly or indirectly affect a company's market demand, cost structure, and financing environment. For example, when GDP growth rate increases, market demand may increase, thereby improving a company's operating performance and credit rating; while changes in the inflation rate may affect a company's costs and product prices, thus impacting its profitability and creditworthiness. Through the vector autoregression model, the impact coefficients of these macroeconomic indicators on corporate credit scores can be quantified, such as obtaining the impact coefficient of GDP growth rate on corporate credit scores. Furthermore, changes in corporate credit scores can be predicted based on changes in macroeconomic indicators. This provides companies with more comprehensive risk warnings, enabling them to adjust their business strategies in advance based on changes in the macroeconomic environment.

[0253] In step ten, the specific method for integrating traditional credit data is as follows:

[0254] First, traditional credit data is screened, including bank loan repayment records (whether repayments were made on time, number of overdue payments, etc.) and commercial credit transaction records (accounts payable payment status, etc.), reflecting the company's historical credit performance.

[0255] Calculate the debt-to-equity ratio (AR). Current ratio CR Return on equity (ROE) Totall liabilties Total represents the company's total liabilities. assets The total assets of the enterprise, Current assets Current assets are the company's current assets. libilities Net is a company's current liability. income Average is the company's net profit. equity This refers to the company's average shareholders' equity.

[0256] Assess a company's financial solvency, liquidity, and profitability. Company size data: Measure company size using indicators such as registered capital, total number of employees, and operating revenue.

[0257] Next, factor analysis is used to reduce the dimensionality of traditional credit data. Let the traditional credit data matrix be Y = (y ij ) n×p Where n is the number of enterprise samples and p is the number of traditional credit indicators;

[0258] After standardizing Y, its correlation coefficient matrix R is calculated;

[0259] The eigenvalues ​​of the correlation coefficient matrix R are obtained through eigenvalue decomposition: λ1≥λ2≥…≥λ p and the corresponding eigenvectors u1, u2, ..., u p ;

[0260] Select the first q feature vectors (so that the cumulative contribution rate reaches a certain threshold, such as 80%) to construct the factor loading matrix A = (u1, u2, ..., u...). q );

[0261] The dimensionality-reduced traditional credit data Z = YA is then concatenated with the innovation score S to form a comprehensive credit assessment data matrix M = (Z|S).

[0262] A credit scoring model is constructed using the random forest algorithm. Random forest is an ensemble learning algorithm composed of multiple decision trees. It constructs multiple decision tree models by sampling the training data with replacement, and finally combines the predictions of multiple decision trees to make a final decision. Random forest can effectively handle high-dimensional data, avoid overfitting, and has good interpretability.

[0263] The comprehensive credit assessment data matrix M is divided into a training set and a test set. On the training set, the parameters of the random forest, such as the number of decision trees (n), are adjusted. e stunatirs), maximum depth (max) d epth), minimum number of sample splits (min) s amples s Methods such as plit are used to find the optimal parameter combination by combining grid search with cross-validation (such as 5-fold cross-validation).

[0264] During cross-validation, the average accuracy, recall, and F1 score of the model under different parameter combinations are calculated, and the parameter combination that achieves the optimal results for these metrics is selected.

[0265] The trained randomized Rylin model was evaluated using a test set.

[0266] The accuracy of the calculation model. Recall in, The evaluation assesses the model's ability to classify enterprises with different credit ratings. TP represents true positives, indicating the number of samples correctly predicted as positive; TN represents true negatives, indicating the number of samples correctly predicted as negative; FP represents false positives, indicating the number of samples incorrectly predicted as positive; and FN represents false negatives, indicating the number of samples incorrectly predicted as negative.

[0267] Simultaneously, the receiver operating characteristic curve (ROC curve) is plotted, and the area under the curve (AUC value) is calculated. The closer the AUC value is to 1, the better the predictive performance of the model.

[0268] The above technical solutions are explained as follows: By screening data such as bank loan repayment records and commercial credit transaction records, the historical credit performance of enterprises can be intuitively reflected. Indicators such as debt-to-equity ratio, current ratio, and return on net assets can comprehensively assess the enterprise's financial solvency, liquidity, and profitability. Enterprise size data such as registered capital can help judge the enterprise's strength, providing rich basic information for credit assessment. Factor analysis dimensionality reduction simplifies many traditional credit indicators, reduces data redundancy, retains key information, improves the efficiency of subsequent analysis, and does not lose important features. The random forest algorithm constructs a credit scoring model, which can effectively handle the integrated high-dimensional data, avoid overfitting, and adjust parameters through grid search combined with cross-validation to find the optimal model configuration and improve model accuracy. Using multiple indicators such as precision, recall, and F1 score to evaluate and draw ROC curves and calculate AUC values ​​can comprehensively measure the model's classification ability and predictive performance for enterprises with different credit ratings, making the assessment results more scientific and reliable.

[0269] In step eleven: the specific method for constructing a credit risk early warning model is as follows:

[0270] The debt-to-equity ratio (AR) and current ratio (CR) are standardized, and the original values ​​are set as x. i The standardized value is Using formula in Let σ be the mean of the indicator, and σ be the standard deviation of the indicator.

[0271] A credit risk early warning model is constructed using a logistic regression model. Let the early warning result be Y (where 0 represents low risk and 1 represents high risk), and the influencing factors be X1, X2, ..., X... n (Including innovation points-related indicators and traditional financial indicators), the logistic regression model formula is: Where β0, β1, ..., β n The model parameters are estimated using the maximum likelihood estimation method;

[0272] Early warning threshold determination: Through analysis of a large amount of historical data, a suitable early warning threshold P0 is determined; when P(Y=1|X1,X2,…,X…) n When )≥P0, a credit risk warning signal is issued.

[0273] The above technical solution is explained as follows: First, key indicators such as the debt-to-equity ratio and current ratio are standardized to eliminate differences in dimensions and numerical ranges, making them comparable and laying the foundation for accurate subsequent analysis. Next, a logistic regression model is used to comprehensively consider numerous influencing factors, including innovation-related indicators and traditional financial indicators, enabling a comprehensive capture of the driving factors of corporate credit risk and assessing the company's credit risk status from multiple dimensions. Maximum likelihood estimation is used to estimate the model parameters, ensuring their accuracy and reliability. Finally, the early warning threshold is determined through the analysis of a large amount of historical data, making the issuance of early warning signals more scientific and reasonable. When the model's calculation results reach or exceed the early warning threshold, a credit risk early warning signal is issued, helping companies to detect potential risks in advance, adjust their business strategies in a timely manner, and reduce risk losses.

[0274] Furthermore, financial institutions will use the credit rating of enterprises under this innovative points-based system as an important reference when approving corporate loans. For example, enterprises with credit ratings of C1 (excellent) and C2 (good) will be granted lower loan interest rates and higher loan amounts; for enterprises with credit ratings of C3 (medium), the loan interest rate and loan amount will be determined after a risk assessment based on specific circumstances; for enterprises with credit ratings of C4 (average) and C5 (poor), a careful decision will be made on whether to grant a loan and the loan terms based on a thorough risk assessment. Let the loan amount calculation formula be L=φ×S total ×A, where L is the loan amount, φ is the loan amount adjustment coefficient (determined based on the financial institution's risk appetite and market conditions), and A is the company's asset size. The loan interest rate calculation formula is r=μ-ν×S total Where r is the loan interest rate, μ is the base interest rate, and ν is the interest rate adjustment coefficient, which is determined according to the policies of financial institutions.

[0275] For example, investment institutions can use a company's innovation competitiveness score (S) as a basis for investment decisions. comp Determine the percentage of investment amount, let's call it I. p The calculation formula is I p =τ×S comp , where τ is the investment ratio adjustment coefficient, which is determined by the investment institution based on its own risk appetite and market conditions.

[0276] Step 12: The specific methods for establishing an enterprise innovation competitiveness assessment model are as follows:

[0277] Based on the innovation score, we further analyze the company's innovation speed, the uniqueness of its innovation, and the sustainability of its innovation;

[0278] The speed of innovation is measured by the frequency, F, of launching new products.

[0279] The uniqueness of an innovation is reflected in the novelty assessment of a patent;

[0280] The sustainability of innovation is reflected by the stability of R&D investment, specifically using the coefficient of variation (CV) of R&D investment. R , Where σ R The standard deviation of R&D investment (The average R&D investment);

[0281] These new indicators will be combined with the various indicators in the innovation score to construct a comprehensive indicator system for evaluating corporate innovation competitiveness;

[0282] The relative weights between the newly included indicators and the original innovation score indicators were determined using the analytic hierarchy process.

[0283] A judgment matrix B is constructed based on seven factors: innovation speed, uniqueness, sustainability, innovation input, innovation results, innovation management, and market influence.

[0284]

[0285] Among them, b ij The meaning of a ij Similarly, it represents the importance scale of factor i relative to factor j;

[0286] By calculating the largest eigenvalue of the matrix and its corresponding eigenvector, and performing a consistency check, the weight vectors of each factor are obtained, as follows:

[0287] and

[0288] Calculate the innovation competitiveness score: Let the enterprise's innovation competitiveness score be S. comp The calculation formula is:

[0289]

[0290] In the previous scheme for constructing an enterprise innovation competitiveness assessment model, CVR represents the coefficient of variation of R&D investment, and its calculation formula is as follows: Where σ R The standard deviation of R&D investment This represents the average R&D investment.

[0291] Formula 1-CVR represents the positive contribution of R&D investment stability to a company's innovation competitiveness, converted using the coefficient of variation of R&D investment. A smaller coefficient of variation indicates more stable R&D investment, resulting in a larger 1-CVR value and a greater contribution to the company's innovation competitiveness. For example, if a company's R&D investment fluctuates very little each year, its CVR value will be smaller, and its 1-CVR value will be larger. Therefore, this component will contribute more significantly to the overall innovation competitiveness score.

[0292] because Standard deviation σ R ≥0, mean (Because R&D investment cannot be 0 or negative, otherwise there would be no R&D investment), therefore CVR ≥ 0. The range of 1-CVR is (-∞, 1). However, in practice, CVR is usually not infinitely large (because although a company's R&D investment fluctuates, it does not fluctuate indefinitely), and CVR is generally greater than 0. Therefore, the actual range of 1-CVR is usually (0, 1). For example, when a company's R&D investment is very stable, CVR approaches 0, and 1-CVR approaches 1; when a company's R&D investment fluctuates greatly, CVR is larger, and 1-CVR will be smaller.

[0293] But it will not be less than 0.

[0294] The above technical solution is explained as follows: A comprehensive evaluation index system is constructed and weights are determined to calculate the innovation competitiveness score, providing multifaceted benefits to enterprises and the market. New product launch frequency measures the speed of innovation, reflecting the enterprise's innovation pace and market responsiveness; patent novelty assesses the uniqueness of innovation, highlighting the enterprise's differentiated advantages; the coefficient of variation of R&D investment measures the sustainability of innovation, reflecting the stability of the enterprise's innovation investment and its long-term development potential. Combining these new indicators with the original innovation score indicators allows for a comprehensive evaluation of the enterprise's innovation competitiveness from multiple dimensions. The analytic hierarchy process (AHP) is used to determine the weights, making the importance of each indicator in the evaluation system more scientific and reasonable. Calculating the innovation competitiveness score provides enterprises with clear and quantitative competitiveness assessment results, enabling them to identify their strengths and weaknesses, optimize innovation strategies, resource allocation, and operational management accordingly, and enhance their innovation competitiveness.

[0295] Although embodiments of the invention have been shown and described, the scope of the invention will be defined by the appended claims and their equivalents by those skilled in the art.

Claims

1. A corporate credit investigation method based on an innovative points system, characterized in that: Specifically, the following steps are included: Step 1: Establish the primary indicators and weights for the innovation points system. The primary indicators include innovation input, innovation results, innovation management, and market influence. Step 2: Refine the primary indicators of the innovation points to obtain secondary indicators; Step 3: Calculate the scores for each secondary indicator; Step 4: Determine the weight of each secondary indicator; Step 5: Calculate the scores for each primary indicator; Step Six: Calculate the overall innovation score; Step 7: Determine credit rating; Step 8: Establish a dynamic adjustment mechanism; Step Nine: Incorporate macroeconomic factors into credit reporting; Step 10: Integrate traditional credit data; Step 11: Construct a credit risk early warning model; Step 12: Establish an enterprise innovation competitiveness assessment model.

2. The enterprise credit investigation method based on an innovation points system according to claim 1, characterized in that, In Step One: The specific method for establishing the primary indicators and weights for the innovation score is as follows: Let the set of primary indicators be I = {I1, I2, I3, I4}, where I1 represents innovation input, I2 represents innovation output, I3 represents innovation management, and I4 represents market influence. The corresponding weight vector is W. 1 And the calculation is as follows: in, The weights are determined using the analytic hierarchy process (AHP), as detailed below: First, construct judgment matrix A. For the four factors—innovation input I1, innovation results I2, innovation management I3, and market influence I4—compare their relative importance pairwise. The judgment matrix is ​​as follows: Among them, a ij The importance scale represents factor i relative to factor j, with values ​​ranging from 1 to 9 and their reciprocals. 1 indicates that both are equally important, 3 indicates that factor i is slightly more important than factor j, 5 indicates that factor i is significantly more important than factor j, 7 indicates that factor i is strongly more important than factor j, and 9 indicates that factor i is extremely more important than factor j. Values ​​2, 4, 6, and 8 are intermediate values. The largest eigenvalue λ of the judgment matrix A is calculated. max And its corresponding eigenvector W, and perform a consistency check; The consistency index (CI) is calculated as follows: Where n is the matrix order, and the random consistency index RI is obtained from the standard table based on the matrix order; Consistency ratio When CR < 0.1, the judgment matrix is ​​considered to have satisfactory consistency, and the resulting eigenvector W is the weight vector W of each factor. 1 .

3. The enterprise credit investigation method based on an innovation points system according to claim 1, characterized in that, Step two involves refining the primary indicators of the innovation score to obtain secondary indicators, including specific details on innovation input, innovation results, innovation management, and market influence. The specific calculation methods are as follows: Innovation investment includes R&D funding and R&D personnel investment; Let R be the company's R&D investment in year t. t The proportion of the annual operating revenue was Among them O t Let t be the company's operating revenue in year t. Let P be the number of R&D personnel in the company, and let T be the proportion of the company's total employees. These innovative achievements include the number of patent applications and sales revenue from new products; Let N be the number of patents filed by the company in year t. a,t The number of patents granted is N g,t The patent conversion rate is r p,t ; Where, when N a,t When r = 0, p,t =0; The company's sales revenue from its new products in year t is S. n,t The proportion of total sales revenue in that year was r. S,n,t The calculation is as follows: Among them, S t Let t be the company's total sales revenue in year t. Innovation management includes the completeness of innovation strategic planning and the effectiveness of innovation incentive systems; The completeness of an enterprise's innovation strategy plan is assessed using a scoring method, with a maximum score of ten. The score for the completeness of an enterprise's innovation strategy plan is denoted as S. s ; The effectiveness of an innovation incentive system is assessed using a scoring method, with a maximum score of ten. The company's innovation incentive system receives an S score for effectiveness. i ; Market influence includes brand awareness and market share. Through market research, calculate the brand awareness rate (B) of the enterprise in the target market. k ; The company's market share in its industry is M. s The calculation is as follows: Where Q represents the sales volume of the company's products or services. total This represents the total sales volume of the industry.

4. The enterprise credit investigation method based on an innovation points system according to claim 1, characterized in that, Step 3: The calculation of secondary indicator scores includes R&D funding investment score, R&D personnel investment score, number of patent applications score, patent conversion rate score, new product sales revenue score, innovation strategy planning completeness score, market share score, brand awareness score, and innovation incentive system effectiveness score. The calculation method for the R&D funding investment score is as follows: Let the scoring function for R&D funding investment be f. R (r R,t ), f R (r R,t )=α1×r R,t Where α1 is the adjustment coefficient, r R,t Score for R&D funding investment; The calculation method for R&D personnel input score is as follows: Let the R&D personnel input score function be f. P (r P ), f P (r P )=α2×r P Where α2 is the adjustment coefficient, r P Scoring should be allocated to R&D personnel; The calculation method for the patent application quantity score is as follows: Let the patent application quantity score function be... Where β1 is the adjustment coefficient, N α,t Score based on the number of patent applications; The patent conversion rate score is calculated as follows: Let the patent conversion rate score function be f. p (r p,t ), f p (r p,t )=β2×r p,t Where β2 is the adjustment coefficient, r p,t The patent conversion rate is scored; The calculation method for the new product sales revenue score is as follows: Let the new product sales revenue score function be... Where γ1 is the adjustment coefficient, r S,n,t Score the sales revenue of new products; The calculation method for the completeness score of the innovation strategy plan is as follows: a scoring method is used, and the score S is... s As a score; The calculation method for the effectiveness score of the innovation incentive system is as follows: a scoring method is used, and the score S is... i As a score; The brand awareness score is calculated as follows: Let the brand awareness score function be f. B (B k ), f B (B k )=δ1×B k Where δ1 is the adjustment coefficient, B k Score brand awareness; The market share score is calculated as follows: Let the market share score function be f. M (M s ), f M (M s )=δ2×M s Where δ2 is the adjustment coefficient, and M is the value of M. s Score based on market share; In step four, the specific method for determining the weights of the secondary indicators is as follows: For each secondary indicator under a primary indicator, the weights are determined using the analytic hierarchy process (AHP). Let the weight vectors for the secondary indicators of innovation input, namely R&D funding input and R&D personnel input, be . and Similarly, construct the judgment matrix and calculate the weights, and determine the weight vectors of the secondary indicators under innovation achievements, innovation management, and market influence. and 5. The enterprise credit investigation method based on an innovation points system according to claim 1, characterized in that, in, In step five: the specific method for calculating the scores of the primary indicators is as follows: The innovation input score is calculated as follows: The score for innovative achievements is calculated as follows: Innovation management score is calculated as follows: Market influence score: Market influence score is calculated as follows: In step six, the specific method for calculating the comprehensive innovation score is as follows: Let the overall innovation score be S. total ,but In step seven, the specific method for credit rating is as follows: Based on the comprehensive score of innovation points S total Credit rating; Let the set of credit ratings be C = {C1, C2, C3, C4, C5}, each corresponding to a different credit rating; In step eight, the specific method for establishing a dynamic adjustment mechanism is as follows: Let the adjustment period be T; At the end of each adjustment period, relevant data from enterprises are collected again, and the overall innovation score is recalculated. Let the adjustment coefficient be θ, then the adjusted innovation score is... The value of θ ranges from (0, 1).

6. The enterprise credit investigation method based on an innovation points system according to claim 1, characterized in that, Step Nine: The specific methods for incorporating macroeconomic factors into credit reporting are as follows: Select macroeconomic indicators closely related to business operations; A vector autoregression model is constructed to analyze the dynamic relationship between macroeconomic indicators and corporate credit indicators. The specific method is as follows: Let the corporate credit score (CS) and GDP growth rate (GDP) be used as the basis for the calculation. g The vector Y consists of the inflation rate IR and the interest rate level R. t =(CS t GDP g,t IR t ,R t ) T ; The expression for the vector autoregressive model is: Where Φ i Let p be the coefficient matrix, and p be the lag order, ∈ t This is the random error term; By estimating the parameters of the vector autoregression model, the influence coefficient of macroeconomic indicators on corporate credit scores is obtained; In step ten, the specific method for integrating traditional credit data is as follows: First, traditional credit data is screened, including bank loan repayment records and commercial credit transaction records; Calculate the debt-to-equity ratio (AR). Current ratio CR Return on equity (ROE) Totall liabilities Total represents the company's total liabilities. assets The total assets of the enterprise, Current assets Current assets are the company's current assets. liabilities Net is a company's current liability. income Average is the company's net profit. equity This refers to the company's average shareholders' equity. Next, factor analysis is used to reduce the dimensionality of traditional credit data. Let the traditional credit data matrix be Y = (y ij ) n×p Where n is the number of enterprise samples and p is the number of traditional credit indicators; After standardizing Y, its correlation coefficient matrix R is calculated; The eigenvalues ​​of the correlation coefficient matrix R are obtained through eigenvalue decomposition: λ1≥λ2≥…≥λ p and the corresponding eigenvectors u1, u2, ..., u p ; Select the first q eigenvectors to construct the factor loading matrix A = (u1, u2, ..., u q ); The dimensionality-reduced traditional credit data Z = YA is then concatenated with the innovation score S to form a comprehensive credit assessment data matrix M = (Z|S). A credit scoring model is constructed using the random forest algorithm. The comprehensive credit assessment data matrix M is divided into a training set and a test set. On the training set, the parameters of the random forest are adjusted, and a grid search combined with cross-validation is used to find a suitable parameter combination. During cross-validation, the average accuracy, recall, and F1 score of the model are calculated as evaluation metrics under different parameter combinations. The trained randomized Rylin model was evaluated using a test set. The accuracy of the calculation model. Recall in, The evaluation assesses the model's ability to classify enterprises with different credit ratings. TP represents true positives, indicating the number of samples correctly predicted as positive; TN represents true negatives, indicating the number of samples correctly predicted as negative; FP represents false positives, indicating the number of samples incorrectly predicted as positive; and FN represents false negatives, indicating the number of samples incorrectly predicted as negative.

7. The enterprise credit investigation method based on an innovation points system according to claim 6, characterized in that, In step eleven: the specific method for constructing a credit risk early warning model is as follows: The debt-to-equity ratio (AR) and current ratio (CR) are standardized, and the original values ​​are set as x. i The standardized value is Using formula in Let σ be the mean of the indicator, and σ be the standard deviation of the indicator. A credit risk early warning model is constructed using a logistic regression model. Let the early warning result be Y (where 0 represents low risk and 1 represents high risk), and the influencing factors be X1, X2, ..., X... n (Including innovation points-related indicators and traditional financial indicators), the logistic regression model formula is as follows: Where β0, β1, ..., β n The model parameters are estimated using the maximum likelihood estimation method. Early warning threshold determination: Through analysis of a large amount of historical data, a suitable early warning threshold P0 is determined; when P(Y=1|X1,X2,…,X…) n When )≥P0, a credit risk warning signal is issued.

8. The enterprise credit investigation method based on an innovation points system according to claim 1, characterized in that, Step 12: The specific methods for establishing an enterprise innovation competitiveness assessment model are as follows: Based on the innovation score, we further analyze the company's innovation speed, the uniqueness of its innovation, and the sustainability of its innovation; The speed of innovation is measured by the frequency, F, of launching new products. The uniqueness of an innovation is reflected in the novelty assessment of a patent; The sustainability of innovation is reflected by the stability of R&D investment, specifically using the coefficient of variation (CV) of R&D investment. R , Where σ R The standard deviation of R&D investment (The average R&D investment); These new indicators will be combined with the various indicators in the innovation score to construct a comprehensive indicator system for evaluating corporate innovation competitiveness; The relative weights between the newly included indicators and the original innovation score indicators were determined using the analytic hierarchy process. A judgment matrix B is constructed based on seven factors: innovation speed, uniqueness, sustainability, innovation investment, innovation results, innovation management, and market influence. Among them, b ij The meaning of a ij Similarly, it represents the importance scale of factor i relative to factor j; By calculating the largest eigenvalue of the matrix and its corresponding eigenvector, and performing a consistency check, the weight vectors of each factor are obtained, as follows: Calculate the innovation competitiveness score: Let the enterprise's innovation competitiveness score be S. comp The calculation formula is:

Images