Enterprise high-quality development evaluation method and system

Abstract

The invention provides an enterprise high-quality development evaluation method and system, and belongs to the technical field of enterprise evaluation, and the method comprises the steps: S1, collecting the multi-dimensional enterprise data of a target enterprise; the multiple dimensions comprise a financial management dimension, a scientific and technological innovation dimension, a talent management dimension, an enterprise associated risk dimension, an enterprise sustainable development dimension and a compliance complaint dimension; s2, establishing a multi-dimensional index library, and performing feature derivation and feature engineering to obtain a feature variable of each dimension; and S3, constructing a dimension scoring model and a comprehensive scoring model, inputting the feature variables of each dimension into the dimension scoring model of the corresponding dimension to obtain a score of each dimension, and inputting the scores of all the dimensions and the feature variables of all the dimensions into the comprehensive scoring model to obtain a high-quality development score of the target enterprise. The method has the beneficial effects that by constructing a multi-dimensional evaluation system, the high-quality development level of an enterprise can be comprehensively evaluated, and the limitation of single-dimensional evaluation is overcome.

Description

Technical Field

[0001] The present invention relates to the field of enterprise evaluation technology, and in particular to a method and system for evaluating high-quality development of an enterprise. Background Art

[0002] With the advancement of global economic integration and increasingly fierce market competition, the high-quality development of enterprises, as the main body of the market economy, has become an important force in promoting economic growth and social progress. The high-quality development of enterprises is no longer limited to the pursuit of maximizing economic benefits, but also emphasizes the comprehensive improvement of enterprises in terms of innovation capabilities, sustainable development, social responsibility and governance structure. In order to achieve these goals, a scientific, objective and fair evaluation system is needed to guide and measure the development quality of enterprises.

[0003] At present, all kinds of enterprises generally face difficulties such as information asymmetry, incomplete data, inaccurate data and even information fraud in the process of enterprise evaluation. Although the traditional enterprise high-quality development evaluation system can provide enterprises with evaluation framework and reference standard guidance to a certain extent, its limitations are becoming increasingly prominent in the face of a rapidly changing economic environment and diversified enterprise development needs. The traditional enterprise high-quality development evaluation system often focuses on financial indicators and market performance, such as profit, revenue, cost, market share and stock price. This single evaluation indicator cannot fully reflect the comprehensive development status of the enterprise. The importance of the enterprise's technological innovation capabilities, talent management and compliance litigation is often overlooked in the traditional enterprise high-quality development evaluation system. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a method and system for evaluating the high-quality development of an enterprise, which can perceive the advantages and disadvantages of enterprise development in multiple dimensions.

[0005] The technical problem solved by the present invention can be achieved by adopting the following technical solutions:

[0006] A method for evaluating high-quality development of an enterprise, comprising:

[0007] Step S1, collecting multi-dimensional enterprise data of the target enterprise, and preprocessing the multi-dimensional enterprise data to obtain preprocessed multi-dimensional enterprise data; wherein the multi-dimensional dimensions include financial management dimension, technological innovation dimension, talent management dimension, enterprise-related risk dimension, enterprise sustainable development dimension, and compliance-related litigation dimension;

[0008] Step S2, establishing a multi-dimensional indicator library according to the pre-processed enterprise data, performing feature derivation and feature engineering according to the atomic indicators in the multi-dimensional indicator library, and obtaining feature variables of each dimension;

[0009] Step S3, construct a dimensional scoring model and a comprehensive scoring model, input the characteristic variables of each dimension into the dimensional scoring model of the corresponding dimension respectively, obtain the score of each dimension, and input the scores of all dimensions and the characteristic variables of all dimensions into the comprehensive scoring model to obtain the comprehensive score of the high-quality development of the target enterprise.

[0010] Preferably, the step S2 comprises:

[0011] Step S201, constructing an indicator system, wherein the indicator system includes an indicator set on which the scoring of each dimension in the multi-dimensionality depends, and the indicator set includes atomic indicators and derived indicators;

[0012] Step S202: performing feature engineering on each indicator item of the indicator system according to the preprocessed enterprise data to obtain feature variables of each indicator item of each dimension.

[0013] Preferably, the index items include positive index, negative index, center index and interval index;

[0014] The step S202 further includes:

[0015] The characteristic variables of each indicator item of each dimension are processed in an orderly manner, and the negative indicator, the center indicator and the interval indicator are converted into a consistent manner according to the positive indicator.

[0016] Preferably, the dimension scoring model and / or the comprehensive scoring model is an expert model, and the method for constructing the expert model includes:

[0017] Step S301, selecting key indicators from the multi-dimensional indicator library according to the importance and relevance of indicator items;

[0018] Step S302, dividing the collected different enterprise data according to preset attributes to obtain enterprise sample sets with different attributes;

[0019] Step S303, filtering out key indicators corresponding to each dimension from the selected key indicators;

[0020] Step S304, for enterprises with different attributes, different binning values ​​and corresponding weights are defined for the key indicators of each dimension to construct the expert model.

[0021] Preferably, the expert model adopts a multiple linear regression model, and the multiple linear regression model is:

[0022]

[0023] Among them, N represents the total number of key indicators screened; i represents the sequence number of the key indicators screened; feat i represents the i-th key indicator; w i Indicates the weight corresponding to the key indicator; Score i Represents the score output by the dimensional scoring model or the comprehensive scoring model.

[0024] Preferably, the expert model uses nonlinear transformation feature binning assignment:

[0025]

[0026] Among them, i represents the key indicator number obtained by screening, i=1,...,N, N represents the total number of key indicators obtained by screening; feat_score i represents the score after the i-th key indicator is mapped to the corresponding interval; w i Indicates the weight corresponding to the key indicator; Score i Represents the score output by the dimensional scoring model or the comprehensive scoring model.

[0027] Preferably, the dimensional scoring model and / or the comprehensive scoring model is a machine learning model, and the method for constructing the machine learning model includes:

[0028] Step S311, defining positive sample enterprises and negative sample enterprises according to the business status of the enterprises; the positive sample enterprises continue to operate within a predetermined time range, and the negative sample enterprises stop operating within a predetermined time range;

[0029] Step S312, dividing the collected data of different enterprises according to preset attributes to obtain enterprise sample sets with different attributes, and dividing them into training sets and test sets;

[0030] Step S313, selecting key indicators corresponding to each dimension from the multi-dimensional indicator library as model parameters;

[0031] Step S314, inputting the training set into the machine learning model for model fitting, and tuning the model parameters;

[0032] Step S315, evaluating the trained model according to the test set to obtain a final machine learning model.

[0033] Preferably, the step S3 further includes:

[0034] The grade of the target enterprise is determined based on the comprehensive score of the target enterprise's high-quality development.

[0035] Preferably, the step S3 further includes:

[0036] The scores of all the dimensions and the comprehensive score of high-quality development of the target enterprise are standardized within the same business partition, and each business partition corresponds to an explainable business partition label.

[0037] The present invention also provides an enterprise high-quality development evaluation system, which is used to implement the above-mentioned enterprise high-quality development evaluation method, including:

[0038] A data collection and preprocessing module is used to collect multi-dimensional enterprise data of a target enterprise and preprocess the multi-dimensional enterprise data to obtain preprocessed multi-dimensional enterprise data;

[0039] A multi-dimensional indicator library construction module, connected to the data collection and preprocessing module, is used to establish a multi-dimensional indicator library according to the preprocessed enterprise data, perform feature derivation and feature engineering according to the atomic indicators in the multi-dimensional indicator library, and obtain feature variables of each dimension;

[0040] An evaluation module is connected to the multi-dimensional indicator library construction module, and is used to construct a dimensional scoring model and a comprehensive scoring model, and input the characteristic variables of each dimension into the dimensional scoring model of the corresponding dimension respectively to obtain a score for each dimension, and input the scores of all dimensions and the characteristic variables of all dimensions into the comprehensive scoring model to obtain a comprehensive score for the high-quality development of the target enterprise.

[0041] The advantages or beneficial effects of the technical solution of the present invention are:

[0042] The present invention constructs a multi-dimensional evaluation system that covers multiple dimensions such as corporate financial management, technological innovation, human resources management, related risks, sustainable development, and compliance-related litigation. It has a wide coverage and can comprehensively and comprehensively evaluate the level of high-quality development of enterprises, overcoming the limitations of single-dimensional evaluation. At the same time, the comprehensive scoring model conducts a comprehensive score for the high-quality development of enterprises based on the scores of each dimension and its characteristic variables, providing a scientific and comprehensive reference for corporate self-diagnosis, strategic planning, and policy formulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 This is a flow chart of a method for evaluating high-quality development of an enterprise in a preferred embodiment of the present invention;

[0044] Figure 2 A schematic diagram of various dimensions and data categories associated with the dimensions in a preferred embodiment of the present invention;

[0045] Figure 3 This is a flow chart of step S2 in a preferred embodiment of the present invention;

[0046] Figure 4 A schematic diagram of a flow chart of a method for constructing an expert model in a preferred embodiment of the present invention;

[0047] Figure 5 A schematic diagram of a flow chart of a method for constructing a machine learning model in a preferred embodiment of the present invention;

[0048] Figure 6 This is a structural block diagram of a system for evaluating the high-quality development of an enterprise in a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0051] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.

[0052] The embodiment of the present invention provides an implementation scheme for quantitative indicators of enterprise development level, namely, an implementation scheme for an enterprise high-quality development evaluation system, which describes a comprehensive score based on multiple dimensions of the enterprise. It is a group-comparable, comprehensive, time-varying and individual-different quantitative indicator. It can improve or avoid certain factors by influencing the indicators included in multiple dimensions associated with the enterprise development score, so as to achieve the effect of promoting the high-quality development of the enterprise, and play a role in comprehensive perception, self-motivation and decision-making assistance.

[0053] See also Figure 1 In a preferred embodiment of the present invention, based on the above problems existing in the prior art, a method for evaluating high-quality development of an enterprise is provided, comprising:

[0054] Step S1, collecting multi-dimensional enterprise data of the target enterprise, and pre-processing the multi-dimensional enterprise data to obtain the pre-processed multi-dimensional enterprise data; wherein the multi-dimensional data includes financial management dimension, technological innovation dimension, talent management dimension, enterprise-related risk dimension, enterprise sustainable development dimension, and compliance-related litigation dimension;

[0055] Step S2, establishing a multi-dimensional indicator library based on the pre-processed enterprise data, performing feature derivation and feature engineering based on the atomic indicators in the multi-dimensional indicator library, and obtaining feature variables of each dimension;

[0056] Step S3, construct a dimensional scoring model and a comprehensive scoring model, input the characteristic variables of each dimension into the dimensional scoring model of the corresponding dimension respectively, obtain the score of each dimension, and input the scores of all dimensions and the characteristic variables of all dimensions into the comprehensive scoring model to obtain the comprehensive score of high-quality development of the target enterprise.

[0057] Specifically, the enterprise high-quality development evaluation method of the embodiment of the present invention includes four stages: data collection and preprocessing, establishment of a multi-dimensional indicator library, model construction, and scoring output. Among them:

[0058] In the data collection and preprocessing stage, enterprise data related to the comprehensive score of high-quality development of the target enterprise is collected; the enterprise data includes but is not limited to data of multiple dimensions related to the high-quality development index score of the target enterprise and basic information data of the enterprise. Multiple dimensions include:

[0059] The financial management dimension is used to reflect the data related to the financial status of the enterprise to be evaluated;

[0060] The technological innovation dimension is used to reflect the data related to the development potential and innovation ability of the enterprise to be evaluated;

[0061] The human resources management dimension is used to reflect data related to personnel turnover and salary levels of the enterprise to be evaluated;

[0062] The enterprise-related risk dimension is used to reflect data related to transactions and potential conflicts of interest of the enterprise to be evaluated;

[0063] The corporate sustainable development dimension is used to reflect data related to the corporate social responsibility, environmental protection and long-term development strategy of the company to be evaluated;

[0064] The compliance and litigation dimension is used to reflect data related to the legal compliance, litigation cases and penalties for violations of the company to be evaluated.

[0065] The scoring of each dimension relies on different indicator libraries to obtain key indicators (or important indicators) that are more relevant to the dimension. Please refer to Figure 2 , illustrates the various dimensions of the comprehensive scoring of high-quality development of an enterprise and the associated data categories in one embodiment. Figure 2 , the data scope of each dimension is described by examples. It is worth mentioning that the data scope of each dimension described below is also exemplary, and is not limited to the following data in actual implementation, and there are differences depending on the different settings of some preconditions described above, specifically:

[0066] The data in the enterprise financial management dimension include but are not limited to the tax data, profit data, asset and liability data, cash flow data and other financial status-related data of the enterprise to be evaluated.

[0067] The dimensions of enterprise technological innovation include but are not limited to key project investment data, existing scientific and technological achievements data, scientific and technological R&D investment data, and other data related to development potential and innovation capabilities of the evaluated enterprises.

[0068] The dimensions of enterprise human resources management include but are not limited to data such as the basic situation of employees, employee turnover, and employee salary levels of the enterprise to be evaluated.

[0069] The enterprise-related risk dimensions include, but are not limited to, shareholder information, foreign investment information, and other related-party transactions and potential conflicts of interest related data of the enterprise to be evaluated.

[0070] The enterprise sustainable development dimension includes, but is not limited to, data such as resource usage, environmental assessment information, and resource utilization of the enterprise to be evaluated.

[0071] In this embodiment, the enterprise to be evaluated is the target enterprise mentioned above. The target enterprise can be a single enterprise to achieve high-quality development evaluation of a single enterprise; the target enterprise can also be multiple enterprises in a specific industry to achieve horizontal comparison and comprehensive evaluation of the overall development trend of the industry.

[0072] Furthermore, after collecting the data of multiple dimensions of the current enterprise, data preprocessing is performed. Data preprocessing includes but is not limited to data standardization, data conversion, deduplication, outlier processing and data verification to improve data quality. The following describes this step in detail in conjunction with this embodiment:

[0073] Step S101, data standardization processing: standardize the collected data, unify the date, time, currency and other formats of the data, eliminate processing obstacles caused by different formats, and provide consistent input for subsequent data analysis and model building.

[0074] Step S102, data conversion: perform necessary conversion on the data to meet the needs of specific analysis or models. The data conversion step includes but is not limited to code conversion, unit conversion, etc.; for example, converting financial data from US dollars to RMB.

[0075] Step S103, deduplication processing: by identifying and deleting duplicate data records, the uniqueness and accuracy of the data are ensured to avoid data redundancy and misleading conclusions in subsequent analysis.

[0076] Step S104, outlier processing: use statistical methods or business rules to identify outliers in the data, and decide subsequent outlier processing operations based on business rules, such as deleting, correcting or retaining outliers, to reduce noise and improve the reliability of data analysis.

[0077] Step S105, data verification: verify the rationality and accuracy of the data through comparison, verification, etc. according to business logic.

[0078] During the establishment phase of the multi-dimensional indicator library, atomic-level indicators are established based on the dimensions and data conditions; feature derivation and feature engineering are then performed based on the atomic-level indicators.

[0079] As a preferred embodiment, wherein Figure 3 As shown, step S2 includes:

[0080] Step S201, constructing an indicator system, wherein the indicator system includes an indicator set on which the scoring of each dimension in the multi-dimensionality depends, and the indicator set includes atomic indicators and derived indicators;

[0081] Each dimension corresponds to an indicator library, and each indicator library consists of an indicator set of the corresponding dimension. The indicator set contains two types of indicators, namely atomic indicators and derived indicators. Atomic indicators are the basis of the indicator system, which are directly derived from enterprise data and have the characteristic of being inseparable. The following are examples of atomic indicators established based on different dimensions:

[0082] Specifically, based on financial management dimensions and data conditions, atomic indicators including but not limited to operating profit, sales, current assets, current liabilities, etc. are established in the financial indicator library.

[0083] Specifically, based on the dimensions and data of scientific and technological innovation, atomic indicators including but not limited to the number of R&D personnel and the number of patent applications are established in the scientific and technological innovation indicator database.

[0084] Specifically, based on the talent management dimensions and data conditions, atomic indicators including but not limited to the number of employees, employee salary levels, and the number of high-tech personnel are established in the human resource indicator database.

[0085] Specifically, based on the enterprise-related risk dimensions and data conditions, atomic indicators including but not limited to the number of related enterprises and the amount of foreign investment are established in the related risk indicator library.

[0086] Specifically, based on the enterprise's sustainable development dimensions and data, atomic indicators including but not limited to carbon emissions and the number of projects passed by environmental impact assessment are established in the enterprise sustainable development index library.

[0087] Specifically, based on the compliance-related litigation dimensions and data, atomic indicators including but not limited to penalty content, penalty time, etc. are established in the compliance-related litigation indicator library.

[0088] Step S202: perform feature engineering on each indicator item of the indicator system according to the preprocessed enterprise data to obtain feature variables of each indicator item in each dimension.

[0089] In addition to atomic indicators, derived indicators can also be constructed according to business needs and analysis purposes. Derived indicators are obtained by mathematical operations, logical judgments or combinations of atomic indicators, and can provide more in-depth and comprehensive information. In this embodiment, each indicator item is characterized through feature derivation and feature engineering. The following are examples of derived indicators derived from different atomic indicators:

[0090] Specifically, based on the atomic-level indicators of the financial indicator library, the DuPont analysis method is used to derive indicators such as operating profit margin and debt-to-asset ratio.

[0091] Specifically, based on the atomic indicators in the scientific and technological innovation indicator library, indicators such as whether a patent application has been made are generated through the One-Hot Encoding method.

[0092] Specifically, based on the atomic indicators in the human resource index library, the enterprise sustainable development index library and the compliance and litigation index library, the time slicing method is used to generate indicators such as the employee turnover rate in the past month, the enterprise carbon emissions in the past month, and the number of administrative penalties in the past three months.

[0093] Specifically, based on the atomic indicators in the associated risk library, derivative indicators such as investment proportion are generated through calculation.

[0094] As a preferred implementation, the index items include positive index, negative index, center index and interval index;

[0095] Step S202 also includes:

[0096] The characteristic variables of each indicator item in each dimension are processed in an orderly manner, and the negative indicators, center indicators and interval indicators are converted into positive indicators in a consistent manner.

[0097] Specifically, the indicator item types are divided into positive indicators, negative indicators, middle indicators and interval indicators. Positive indicators directly reflect the advantages of the enterprise. The larger the indicator value, the better the enterprise performs in this aspect, such as operating profit, sales, etc. On the contrary, the smaller the indicator value of the negative indicator, the better the enterprise performs in this aspect, such as carbon emissions, number of legal penalties, etc. It is best for the indicator value of the middle indicator to be maintained within a specific range, and the performance is poor if it deviates from this range, such as employee turnover rate, inventory turnover rate, etc. The indicator value of the interval indicator is within an acceptable range within a specific range, and exceeding this range may indicate potential risks, such as debt-to-asset ratio, credit rating, etc.

[0098] In this embodiment, when characterizing each indicator item, the indicator characteristics are ordered according to the principle that the larger the enterprise's high-quality score and dimension score, the better the development level in the comparable group. When ordering the indicator characteristics of each dimension, the negative indicators, the middle indicators and the interval indicators are transformed into consistent positive indicators, so that all indicators can be compared and scored under the same standard, so as to build a unified and comparable high-quality development evaluation indicator system.

[0099] For the ordered processing of indicator characteristics, operations such as sorting, segmenting or mapping indicator values ​​to new scoring intervals are usually adopted to ensure that all indicators can be scored and compared in the same direction (i.e. positive).

[0100] For negative indicators, convert them into positive indicators. This is usually achieved by taking negative values, reciprocals or other mathematical transformations, so that the larger the converted indicator value, the better the company's performance in this aspect. For example, carbon emissions can be converted into "carbon emission reduction" or "carbon emission efficiency" (such as carbon emissions per unit of output value), thereby converting it into a positive indicator.

[0101] For the conversion of the centered indicator, a suitable center point or optimal interval is found, and then the deviation between the actual value and the center point is calculated. The smaller the deviation, the better the company performs in this aspect. Specific conversion methods may include calculating the absolute value deviation, square deviation, or using piecewise functions.

[0102] For interval indicator conversion, first determine the acceptable range or target interval. For indicator values ​​beyond the interval, they can be converted through a penalty function (such as linear decreasing, exponential decreasing, etc.), so that the part beyond the interval is negatively affected in the score. At the same time, for indicator values ​​within the interval, their original values ​​can be maintained or appropriately weighted.

[0103] The comprehensive score for high-quality development of enterprises in the present invention is composed of the enterprise financial management score, enterprise scientific and technological innovation score, enterprise human resources management score, enterprise compliance and litigation score, enterprise related risk score, and enterprise sustainable development score. It has a wide coverage. By establishing a multi-dimensional evaluation system, enterprises can be comprehensively scored, avoiding the limitations of single-dimensional evaluation.

[0104] In the model building stage, based on expert knowledge methods and machine learning methods, samples are defined, features are screened, weights are defined, model fitting is performed, and parameters are optimized to build scoring models for each single dimension and a comprehensive scoring model for high-quality development of enterprises. Model building includes machine learning models and expert models, among which:

[0105] The expert model is an initial reliable indicator set formed by enterprise development analysis experts according to the steps, and the weight calculation algorithm is used to determine the weight of each indicator, and then the weighted algorithm is used to obtain the dimension scoring model and the comprehensive scoring model of enterprise high-quality development. As a preferred embodiment, the dimension scoring model and / or the comprehensive scoring model are expert models, such as Figure 4 As shown in Figure 2, the method for constructing the expert model includes:

[0106] Step S301, selecting key indicators from a multi-dimensional indicator library according to the importance and relevance of indicator items;

[0107] Specifically, based on expert experience, important features are selected from the indicator library of each dimension mentioned above according to the importance and relevance of each indicator item as key indicators and as input parameters of the model;

[0108] Step S302, dividing the collected different enterprise data according to preset attributes to obtain enterprise sample sets with different attributes;

[0109] Among them, the preset attributes may include but are not limited to the different industry categories, business scale, and value-added tax taxpayer type of the enterprise.

[0110] Step S303, filtering out key indicators corresponding to each dimension from the selected key indicators;

[0111] Among them, the key indicators corresponding to each dimension are further screened out from the key indicators selected in step S301. The scoring indicators of the dimension scoring model of each dimension are different, and they are emphasized according to different preconditions. On the basis of the feature variables formed by each dimension, the combination of preconditions is superimposed, and the feature variables are reselected, and the corresponding dimension scoring model is trained on the selected feature variable set. The following are examples of key indicators screened out in different dimensions:

[0112] The enterprise financial management scoring model selects DuPont analysis financial indicators, the enterprise scientific and technological innovation scoring model selects important indicators from the enterprise scientific and technological innovation indicator library, the enterprise human resources management scoring model selects important indicators from the enterprise human resources indicator library, the enterprise compliance and litigation scoring model selects important indicators from the enterprise compliance indicator library, the enterprise related risk scoring model selects important indicators from the enterprise related risk indicator library, the enterprise related risk scoring model selects important indicators from the enterprise related risk indicator library, and the enterprise sustainable development scoring model selects important indicators from the enterprise sustainable development indicator library.

[0113] Step S304, for enterprises with different attributes, different binning values ​​and corresponding weights are defined for key indicators of each dimension to construct an expert model.

[0114] Specifically, after statistical analysis of enterprises with different attributes, such as different industries and different sizes, different bin values ​​and corresponding weights are defined for the key indicators of each dimension.

[0115] As a preferred implementation, the expert model adopts a multiple linear regression model, and the multiple linear regression model is:

[0116]

[0117] Where N represents the total number of key indicators screened; i represents the sequence number of the key indicators screened, i = 1, ..., N, N is at least not less than 3; feat i represents the i-th key indicator; w i Indicates the weight corresponding to the key indicator; Score i Represents the score output by the dimensional scoring model or the comprehensive scoring model.

[0118] As a preferred implementation, the expert model uses nonlinear transformation feature binning assignment, such as equal-width binning, equal-frequency binning, cluster binning, minimum entropy method, etc.:

[0119]

[0120] Among them, i represents the key indicator number obtained by screening, i=1,...,N, N represents the total number of key indicators obtained by screening; feat_score i represents the score after the i-th key indicator is mapped to the corresponding interval; w i Indicates the weight corresponding to the key indicator; Score i Represents the score output by the dimensional scoring model or the comprehensive scoring model.

[0121] In this embodiment, for each model, the selected features include at least three. For example, for the enterprise financial management scoring model, the key indicators for the input parameters may be three or more.

[0122] The machine learning model, based on the characterized data, divides the characterized data into positive samples and negative samples. The positive sample enterprises continue to operate within a limited time range, and the negative sample enterprises terminate their operations within a limited time range. Then, the dimension scoring model of each dimension and the comprehensive scoring model of enterprise high-quality development are trained based on the data of each dimension. As a preferred embodiment, the dimension scoring model and / or the comprehensive scoring model are machine learning models, such as Figure 5 As shown, the method for building a machine learning model includes:

[0123] Step S311, defining positive sample enterprises and negative sample enterprises according to the business status of the enterprises; the positive sample enterprises continue to operate within a predetermined time range, and the negative sample enterprises stop operating within a predetermined time range;

[0124] Step S312, dividing the collected data of different enterprises according to preset attributes to obtain enterprise sample sets with different attributes, and dividing them into training sets and test sets;

[0125] In this embodiment, the preset attributes may include, but are not limited to, the different industry categories of the enterprise, the scale of operation, and the type of value-added tax taxpayer.

[0126] The training set and the test set are divided according to a certain ratio. For example, the samples are divided into the test set and the training set according to a sampling ratio of 2:8.

[0127] Step S313, selecting key indicators corresponding to each dimension from the multi-dimensional indicator library as model parameters;

[0128] In this embodiment, the key indicators can be obtained by performing the first step of preliminary screening from the features in the indicator library through methods such as null value rate, correlation, feature importance, and principal component analysis (PCA).

[0129] Step S314, inputting the training set into the machine learning model, performing model fitting, and tuning by adjusting the model parameters;

[0130] In this embodiment, the machine learning model can be constructed using a multivariate linear regression model and a distributed gradient boosting library (XGBoost) model.

[0131] Step S315, evaluate the trained model according to the test set to obtain the final machine learning model.

[0132] In this embodiment, the accuracy of the model can be evaluated based on the confusion matrix. The elements in the confusion matrix include true positives (TP), false positives (FP), true negatives (TN) and false negatives (FN). In each confusion matrix, the recall rate, precision rate and accuracy rate can be calculated. Among them, the recall rate TPR = TP / (TP+FN), the precision rate FPR = FP / (FP+TN), and the accuracy rate ACC = (TP+TN) / (TP+TN+FN+FP).

[0133] The ROC curve uses the precision rate FPR as the x-axis and the recall rate TPR as the y-axis. By solving the area under the ROC curve and the coordinate axis, the AUC (Area under Curve) is used to intuitively evaluate the quality of the classifier. The AUC value is between 0.5 and 1. The larger the AUC value, the better the performance of the classifier.

[0134] In the scoring output stage, the characteristic variables of each dimension in the multiple dimensions are input into the corresponding dimension scoring model of each dimension to obtain the scoring value of each dimension in the multiple dimensions; then, the score of each dimension in the multiple dimensions and the characteristic variables of each dimension are input into the comprehensive scoring model of high-quality development of the enterprise to obtain the comprehensive scoring score of the high-quality development of the target enterprise.

[0135] In this embodiment, two enterprise evaluation methods are provided, including steps S301 to S304 for expert knowledge model scoring method and steps S311 to S315 for machine learning model scoring method. By combining expert knowledge and machine learning, feature variables and weights are dynamically adjusted according to preconditions to ensure the accuracy and adaptability of the scoring model, thereby providing a more personalized and accurate enterprise development evaluation.

[0136] As a preferred implementation, step S3 further includes:

[0137] The grade of the target enterprise is determined based on the comprehensive score of its high-quality development.

[0138] Specifically, the target enterprise's grade is directly determined based on its comprehensive score of high-quality development. Specifically, by setting enterprise grading standards, similar enterprises are ranked from high to low according to their comprehensive scores and divided into different grades. Enterprise grading standards can be set and flexibly adjusted according to actual conditions;

[0139] As a preferred implementation, step S3 further includes:

[0140] The scores of all dimensions and the comprehensive score of high-quality development of the target enterprise are standardized into the same business segment, and each business segment corresponds to an explainable business segment label.

[0141] Specifically, in this embodiment, each score and each dimension score in the comprehensive score of the enterprise's high-quality development can be standardized. The score standardization is based on the business partition division standard of the business system of the target enterprise's high-quality development score. The score of the target enterprise's high-quality development score and the score of the dimension are standardized to the corresponding business partition interval, eliminating the deviation between different dimensions caused by differences in scoring standards, ensuring that all scores can be compared and analyzed under the same standard, making the scoring results more fair and objective.

[0142] Each business segment is assigned an interpretable business segment label, which directly corresponds to the performance level of the enterprise in the corresponding dimension, making the scoring results more intuitive and easy to understand, and facilitating users to understand and analyze the performance of the enterprise in different dimensions.

[0143] Exemplarily, this embodiment may output the scoring levels as grades A to F according to the interval definition method shown in Table 1.

[0144] Table 1 Definition method of each scoring level interval

[0145]

[0146]

[0147] As shown in Table 1 above, the scores of the six dimensions are weighted and combined to obtain a comprehensive score. In order to facilitate users to understand the score, the comprehensive score is finally mapped to a score of 1,000 (0 to 1,000).

[0148] See also Figure 6 The present invention also provides a system for evaluating high-quality development of an enterprise, which is used to implement the above-mentioned method for evaluating high-quality development of an enterprise, including:

[0149] The data collection and preprocessing module 1 is used to collect the multi-dimensional enterprise data of the target enterprise and preprocess the multi-dimensional enterprise data to obtain the preprocessed multi-dimensional enterprise data;

[0150] The multi-dimensional indicator library construction module 2 is connected to the data collection and preprocessing module 1, and is used to establish a multi-dimensional indicator library based on the preprocessed enterprise data, and perform feature derivation and feature engineering based on the atomic indicators in the multi-dimensional indicator library to obtain feature variables of each dimension;

[0151] Evaluation module 3 is connected to multi-dimensional indicator library construction module 2, and is used to construct a dimensional scoring model and a comprehensive scoring model. The characteristic variables of each dimension are respectively input into the dimensional scoring model of the corresponding dimension to obtain the score of each dimension, and the scores of all dimensions and the characteristic variables of all dimensions are input into the comprehensive scoring model to obtain the comprehensive score of high-quality development of the target enterprise.

[0152] Furthermore, the system may also include: a model building module, which is used to build a multi-dimensional scoring model and a comprehensive scoring model for high-quality development of enterprises based on expert knowledge methods and machine learning methods.

[0153] Furthermore, it may also include:

[0154] The report output module is used to convert the scoring values ​​of each dimension obtained by the model and the comprehensive scoring results of high-quality development into an analysis report for the target enterprise. The report includes but is not limited to financial data analysis, tax data analysis, etc., and supports multiple formats such as PDF and WORD, and can be customized according to enterprise needs.

[0155] Furthermore, it may also include:

[0156] Storage module, used to store various data and results in the system;

[0157] The query module is used to query the enterprise's various scoring data and the enterprise's scoring results;

[0158] The pre-rule module is used to call the corresponding statistical methods according to the rules based on different pre-conditions to obtain characteristic variables, calculate the scores of each dimension and the enterprise's high-quality development index, so as to ensure the flexibility and adaptability of the evaluation process.

[0159] The standardization module is used to standardize the enterprise high-quality development index score obtained by the enterprise dimension scoring model and the dimension scores obtained by each dimension scoring model into a unified business segment, and provide interpretable business segment labels. This module ensures the comparability and consistency of each score, helping users to more intuitively understand and analyze the performance of enterprises in different dimensions.

[0160] The index interpretation module is used to provide feedback on the development of enterprises from the perspective of scores and indicators according to preset rules, and interpret them to target enterprises. This module helps users understand the key factors affecting the high-quality development of enterprises, provides auxiliary management guidance programs, and formulates incentives to promote the high-quality development of enterprises.

[0161] The method and system finally obtained in the present invention can obtain comprehensive data information related to the enterprise on a convenient and fast basis, and make scientific and reasonable weight allocation for the data information, so as to obtain more valuable and closer to the actual enterprise high-quality development evaluation results.

[0162] Those skilled in the art will appreciate that various aspects of the present invention, or possible implementations of various aspects, may be specifically implemented as systems, methods, or computer program products. Therefore, various aspects of the present invention, or possible implementations of various aspects, may take the form of complete hardware embodiments, complete software embodiments (including firmware, resident software, etc.), or embodiments of combined software and hardware aspects, all collectively referred to herein as "circuits," "modules," or "systems." In addition, various aspects of the present invention, or possible implementations of various aspects, may take the form of computer program products, which refer to computer instructions stored in a memory.

[0163] The memory may be a computer-readable signal medium or a computer-readable storage medium. Computer-readable storage media include but are not limited to electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or apparatuses, or any suitable combination of the foregoing, such as random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable read-only memory (CD-ROM).

[0164] The processor in the computer reads the computer instructions stored in the memory, so that the processor can execute the functional actions specified in each step or the combination of steps in the flowchart; and generate a device for implementing the functional actions specified in each block or the combination of blocks in the block diagram.

[0165] It should be understood that the processor in the computer can be understood as one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers (MCUs), microprocessors, or other electronic components implemented to execute the aforementioned computer instructions.

[0166] Computer instructions can be executed completely on the user's local computer, partially on the user's local computer, as a separate software package, partially on the user's local computer and partially on a remote computer, or completely on a remote computer or server. It should also be noted that in some alternative embodiments, the functions noted in each step in the flow chart or each block in the block diagram may not occur in the order noted in the figure. For example, depending on the functions involved, two steps or two blocks shown in succession may actually be executed roughly simultaneously, or these blocks may sometimes be executed in reverse order.

[0167] The advantages or beneficial effects of adopting the above technical solution are: the present invention constructs a multi-dimensional evaluation system, covering multiple dimensions such as corporate financial management, scientific and technological innovation, human resources management, related risks, sustainable development, and compliance-related litigation. It has a wide coverage and can comprehensively and comprehensively evaluate the high-quality development level of the enterprise, overcoming the limitations of single-dimensional evaluation; at the same time, the comprehensive scoring model conducts a comprehensive scoring of the high-quality development of the enterprise according to the scores of each dimension and its characteristic variables, providing a scientific and comprehensive reference basis for corporate self-diagnosis, strategic planning and policy formulation.

[0168] The above are only preferred embodiments of the present invention, and are not intended to limit the implementation methods and protection scope of the present invention. Those skilled in the art should be aware that all solutions obtained by equivalent substitutions and obvious changes made using the contents of this specification and illustrations should be included in the protection scope of the present invention.

Claims

1. A method for evaluating high-quality development of an enterprise, characterized in that: include: Step S1, collecting multi-dimensional enterprise data of the target enterprise, and preprocessing the multi-dimensional enterprise data to obtain preprocessed multi-dimensional enterprise data; wherein the multi-dimensional dimensions include financial management dimension, technological innovation dimension, human resources management dimension, enterprise-related risk dimension, enterprise sustainable development dimension, and compliance-related litigation dimension; Step S2, establishing a multi-dimensional indicator library according to the pre-processed enterprise data, performing feature derivation and feature engineering according to the atomic indicators in the multi-dimensional indicator library, and obtaining feature variables of each dimension; Step S3, construct a dimensional scoring model and a comprehensive scoring model, input the characteristic variables of each dimension into the dimensional scoring model of the corresponding dimension respectively, obtain the score of each dimension, and input the scores of all dimensions and the characteristic variables of all dimensions into the comprehensive scoring model to obtain the comprehensive score of the high-quality development of the target enterprise.

2. The enterprise high-quality development evaluation method according to claim 1 is characterized in that: The step S2 comprises: Step S201, constructing an indicator system, wherein the indicator system includes an indicator set on which the scoring of each dimension in the multi-dimensionality depends, and the indicator set includes atomic indicators and derived indicators; Step S202: performing feature engineering on each indicator item of the indicator system according to the preprocessed enterprise data to obtain feature variables of each indicator item of each dimension.

3. The enterprise high-quality development evaluation method according to claim 3 is characterized in that: The index items include positive index, negative index, center index and interval index; The step S202 further includes: The characteristic variables of each indicator item of each dimension are processed in an orderly manner, and the negative indicator, the center indicator and the interval indicator are converted into a consistent manner according to the positive indicator.

4. The enterprise high-quality development evaluation method according to claim 1 is characterized in that: The dimension scoring model and / or the comprehensive scoring model is an expert model, and the method for constructing the expert model includes: Step S301, selecting key indicators from the multi-dimensional indicator library according to the importance and relevance of indicator items; Step S302, dividing the collected different enterprise data according to preset attributes to obtain enterprise sample sets with different attributes; Step S303, filtering out key indicators corresponding to each dimension from the selected key indicators; Step S304, for enterprises with different attributes, different binning values ​​and corresponding weights are defined for the key indicators of each dimension to construct the expert model.

5. The enterprise high-quality development evaluation method according to claim 4 is characterized in that: The expert model adopts a multiple linear regression model, and the multiple linear regression model is: Among them, N represents the total number of key indicators screened; i represents the sequence number of the key indicators screened; feat i represents the i-th key indicator; w i Indicates the weight corresponding to the key indicator; Score i Represents the score output by the dimensional scoring model or the comprehensive scoring model.

6. The method for evaluating high-quality development of an enterprise according to claim 4 is characterized in that: The expert model uses nonlinear transformation feature bin assignment: Among them, i represents the key indicator number obtained by screening, i=1,...,N, N represents the total number of key indicators obtained by screening; feat_score i represents the score after the i-th key indicator is mapped to the corresponding interval; w i Indicates the weight corresponding to the key indicator; Score i Represents the score output by the dimensional scoring model or the comprehensive scoring model.

7. The enterprise high-quality development evaluation method according to claim 1 is characterized in that: The dimension scoring model and / or the comprehensive scoring model is a machine learning model, and the method for constructing the machine learning model includes: Step S311, defining positive sample enterprises and negative sample enterprises according to the business status of the enterprises; the positive sample enterprises continue to operate within a predetermined time range, and the negative sample enterprises stop operating within a predetermined time range; Step S312, dividing the collected data of different enterprises according to preset attributes to obtain enterprise sample sets with different attributes, and dividing them into training sets and test sets; Step S313, selecting key indicators corresponding to each dimension from the multi-dimensional indicator library as model parameters; Step S314, inputting the training set into the machine learning model for model fitting, and tuning the model parameters; Step S315, evaluating the trained model according to the test set to obtain a final machine learning model.

8. The method for evaluating high-quality development of an enterprise according to claim 1 is characterized in that: The step S3 further comprises: The grade of the target enterprise is determined based on the comprehensive score of the target enterprise's high-quality development.

9. The method for evaluating high-quality development of an enterprise according to claim 1 is characterized in that: The step S3 further comprises: The scores of all the dimensions and the comprehensive score of high-quality development of the target enterprise are standardized within the same business partition, and each business partition corresponds to an explainable business partition label.

10. A high-quality development evaluation system for enterprises, characterized in that: Used to implement the enterprise high-quality development evaluation method as described in any one of claims 1 to 9, comprising: A data collection and preprocessing module is used to collect multi-dimensional enterprise data of the target enterprise and preprocess the multi-dimensional enterprise data to obtain preprocessed multi-dimensional enterprise data; wherein the multi-dimensional dimensions include financial management dimension, technological innovation dimension, human resources management dimension, enterprise-related risk dimension, enterprise sustainable development dimension, and compliance-related litigation dimension; A multi-dimensional indicator library construction module, connected to the data collection and preprocessing module, is used to establish a multi-dimensional indicator library according to the preprocessed enterprise data, perform feature derivation and feature engineering according to the atomic indicators in the multi-dimensional indicator library, and obtain feature variables of each dimension; An evaluation module is connected to the multi-dimensional indicator library construction module, and is used to construct a dimensional scoring model and a comprehensive scoring model, and input the characteristic variables of each dimension into the dimensional scoring model of the corresponding dimension respectively to obtain a score for each dimension, and input the scores of all dimensions and the characteristic variables of all dimensions into the comprehensive scoring model to obtain a comprehensive score for the high-quality development of the target enterprise.

Images