CN116645014A

CN116645014A - Provider supply data model construction method based on artificial intelligence

Info

Publication number: CN116645014A
Application number: CN202310930411.XA
Authority: CN
Inventors: 彭剑兵; 黄睿; 刘洋; 游奕凡
Original assignee: Hunan Valin E Commerce Co ltd
Current assignee: Hunan Valin E Commerce Co ltd
Priority date: 2023-07-27
Filing date: 2023-07-27
Publication date: 2023-08-25

Abstract

The invention relates to the technical field of intelligent provider data evaluation, in particular to a provider supply data model construction method based on artificial intelligence. The method comprises the following steps: acquiring supplier supply data according to purchasing requirements; carrying out data preprocessing on the supplier supply data to generate standard supply data; performing evaluation division processing on the standard supply data according to a preset evaluation dimension to generate supplier dimension evaluation data; performing Logistic regression on the supplier dimension evaluation data by using a Logistic regression algorithm to generate a standard supply model; importing the supplier dimension evaluation data into a standard supply model to carry out supply prediction processing, and generating a supply evaluation score value; and carrying out score rating mapping processing on the supplied evaluation score value through a preset provider grade to generate a provider score rating. The invention realizes the artificial intelligence supplier data model construction method by carrying out model construction and decision analysis on the supplier supply data.

Description

Provider supply data model construction method based on artificial intelligence

Technical Field

The invention relates to the technical field of intelligent provider data evaluation, in particular to a provider supply data model construction method based on artificial intelligence.

Background

Vendor assessment is a method of assessing vendor performance and risk to ensure efficient operation and sustainable development of the supply chain. The supplier evaluation can cover various aspects of purchasing cost, delivery time, quality, safety, environmental protection, social responsibility and the like, is beneficial to enterprises to know actual performances and problems of the suppliers, and takes corresponding measures to ensure or improve the stability and the efficiency of the supply chain. In recent years, supplier evaluation has become an increasingly important ring due to globalization, increased competition, and concerns of enterprises about supply chain risks. By establishing a scientific supplier evaluation system, potential supply chain risks can be avoided, performance and efficiency of suppliers are improved, purchasing period is shortened, purchasing cost is reduced, and accuracy of purchasing decisions is improved. The conventional supplier evaluation still has a plurality of problems, such as error of evaluation conclusion caused by inconsistent dimensions and indexes, larger gap between the evaluation result and actual conditions caused by insufficient precision of an evaluation method and a model, and high labor cost.

Disclosure of Invention

Based on this, there is a need to provide a vendor provisioning data model construction method based on artificial intelligence to solve at least one of the above-mentioned technical problems.

To achieve the above object, a provider supply data model construction method based on artificial intelligence, the method comprising the steps of:

step S1: acquiring supplier supply data according to purchasing requirements; carrying out data preprocessing on the supplier supply data to generate standard supply data;

step S2: performing evaluation division processing on the standard supply data according to a preset evaluation dimension to generate supplier dimension evaluation data;

step S3: performing Logistic regression on the supplier dimension evaluation data by using a Logistic regression algorithm to generate a standard supply model; importing the supplier dimension evaluation data into a standard supply model to carry out supply prediction processing, and generating a supply evaluation score value;

step S4: carrying out score rating mapping processing on the supplied evaluation score value through a preset provider grade to generate provider score rating;

step S5: and carrying out hierarchical decision processing on the score ratings of the suppliers by utilizing the decision tree model so as to generate a decision scheme of the suppliers.

According to the invention, the supplier supply data is obtained and subjected to data preprocessing, so that noise and errors in the data can be reduced, the accuracy and reliability of the data are improved, the data processing cost is reduced, the data processing time is shortened, the data consistency and comparability are improved, and standard supply data are generated; the standard supply data is subjected to evaluation division processing according to the preset evaluation dimension, so that supplier dimension evaluation data is generated, the comprehensiveness and accuracy of evaluation can be improved, self-improvement of suppliers is promoted, and supplier selection and management strategies are optimized; the Logistic regression algorithm is used for carrying out Logistic regression processing on the provider dimension evaluation data, a supply evaluation model can be established, and the performance of the provider is accurately measured, so that the accuracy and the reliability of the provider evaluation result are improved, the evaluation process is standardized, the influence of artificial subjective factors on the evaluation result is reduced, the possibility of difference of the evaluation result is reduced, the accuracy and the fairness of the provider evaluation result are improved, and the continuous improvement and improvement of the provider in the aspects of quality, cost, delivery period and the like are promoted; the score rating mapping processing is carried out on the supply evaluation score value through the preset supplier grade, so that the supplier score rating is generated, the evaluation result is more visual and easy to understand, the risk is reduced, the stability of a supply chain is improved, the competition consciousness and innovation enthusiasm of suppliers are stimulated, the performance and the service quality are further improved, the enterprise is facilitated to optimize purchasing decisions, and the purchasing efficiency and the supply chain management effect are improved; and grading decision processing is carried out on the score ratings of the suppliers by utilizing a decision tree model, so that a decision result of the suppliers is generated, and classification management can be carried out on the grading data of the suppliers rapidly, so that the decision efficiency is improved, risks in purchasing decision and supply chain management are reduced, all parties in a supply chain are better managed, the supply chain management effect is improved, purchasing flows are standardized, and the management cost is reduced. Therefore, the artificial intelligence provider data model construction method provided by the invention carries out provider grade grading on provider data, and establishes a provider evaluation model by using a Logistic regression algorithm so as to realize the decision support for the admission of provider projects, improve the model precision and reduce the labor cost.

Preferably, step S1 comprises the steps of:

step S11: performing data missing value filling processing on the supplier supply data to generate supply filling data;

step S12: performing outlier detection processing on the supply filling data, and performing logarithmic transformation processing on the supply filling data when abnormal outliers exist, so as to generate supply outlier data;

step S13: denoising the supply outlier data to generate supply denoising data; and carrying out data normalization processing on the supply denoising data to generate standard supply data.

The invention generates the supply filling data by carrying out data missing value filling processing on the supply data of the supplier, fills the missing value in the data by the data missing value filling processing, ensures the data to be more complete, improves the integrity of the data, ensures the reliability of the evaluation result and reduces the data processing cost; the method comprises the steps of performing outlier detection processing on supply filling data, performing logarithmic transformation processing on the supply filling data to generate supply outlier data when abnormal outliers exist, and finding and processing the abnormal outliers through the outlier detection processing on the supply filling data, so that the data quality is improved, external interference factors are eliminated, errors of an evaluation result are reduced, the accuracy of the evaluation result is improved, the reliability and the precision of the data are improved, the value range of the data can be adjusted to be in a smaller range through logarithmic transformation, the influence of the outliers on the evaluation result is reduced, and the method is simple to operate and easy to interpret; denoising the supply outlier data to generate supply denoising data; the data normalization processing is carried out on the supplied denoising data to generate standard supplied data, outliers and noise can be removed through the data denoising processing, so that the quality of the data is improved, the data is more accurate, real and reliable, unnecessary redundant information in the data can be removed, the data is simplified and normalized, the visualization effect of the data is optimized, and the data processing efficiency is improved.

Preferably, step S2 comprises the steps of:

step S21: obtaining supplier evaluation dimension data through supplier demand analysis, and performing supply weight calculation processing on the supplier evaluation dimension data according to a supplier dimension evaluation weight formula to generate a supply evaluation weight coefficient value;

step S22: performing evaluation weight comparison processing on the standard supply data and the supply evaluation weight coefficient value to generate supply weight comparison data;

step S23: performing weight comparison processing on the supply weight comparison data and preset standard supply weight comparison data; dividing the supplier main evaluation dimension into supplier main evaluation dimensions when the supply weight comparison data is larger than the standard supply weight, thereby generating supplier main dimension evaluation data; dividing into vendor minor dimension evaluation dimensions when the provisioning weight contrast data is less than the standard provisioning weight, thereby generating vendor minor dimension evaluation data;

step S24: and performing evaluation dimension sorting processing on the primary dimension evaluation data and the secondary dimension evaluation data of the suppliers to generate dimension evaluation data of the suppliers.

According to the invention, the supplier evaluation dimension data is obtained through the demand analysis of the suppliers, the supplier evaluation dimension data is subjected to the supply weight calculation processing according to the supplier dimension evaluation weight formula, the supply evaluation weight coefficient value is generated, the objectivity of the evaluation result can be improved, the evaluation standard is clear, the decision effect is improved, and the evaluation flow is standardized; performing evaluation weight comparison processing on the standard supply data and the supply evaluation weight coefficient value to generate supply weight comparison data, and establishing and optimizing an evaluation index system so as to improve evaluation accuracy, facilitate subsequent better decision making and selection of a more suitable supplier, and improve data interpretation so as to facilitate subsequent operation of data; performing weight comparison processing on the supply weight comparison data and preset standard supply weight comparison data; dividing the supplier main evaluation dimension into supplier main evaluation dimensions when the supply weight comparison data is larger than the standard supply weight, thereby generating supplier main dimension evaluation data; when the supply weight comparison data is smaller than the standard supply weight, dividing the supply weight comparison data into provider secondary evaluation dimensions so as to generate provider secondary dimension evaluation data, and reasonably evaluating the provider data in different dimensions so as to improve the accuracy and effectiveness of evaluation and provide useful information for subsequent decisions; and carrying out evaluation dimension sorting processing on the primary dimension evaluation data and the secondary dimension evaluation data of the suppliers to generate the dimension evaluation data of the suppliers, so that the data of each supplier in each evaluation dimension can be clearly known, and the evaluation result is quantized, thereby realizing the visualization of the bottom data.

Preferably, the vendor dimension evaluation weight formula in step S21 is specifically as follows:

；

in the method, in the process of the invention,representing vendor dimension rating weight, +.>Indicating the level of quality of the product supplied by the supplier, < >>Representing the response speed of the supplier to supply the product, +.>Indicating the accuracy of delivery of the product supplied by the supplier, < >>Representing the cost of the product supplied by the supplier, < >>Representing the minimum consideration condition number of the supplier evaluation, < ->Representing the maximum consideration condition number of the supplier evaluation, < ->Representing the supplier company size factor,/-, for>Representing the after-sales quality of service coefficient of the provider product, < >>Representing the supplier product supply stability factor, +.>Representing the total score contribution value of the supplier supply, < ->Representing the vendor dimension rating weight modifier.

The invention constructs a supplier dimension evaluation weight formula which fully considers the quality level of the products supplied by suppliersResponse speed of vendor supplied products>Delivery accuracy of products supplied by suppliers->Cost of product supplied by suppliers->Supplier evaluation minimum consideration condition number +.>Maximum supplier ratingConsider condition number->Vendor company Scale factor->After-sales service quality factor of vendor product>Stability factor of the supplier's product supply >Total score contribution value supplied by supplier +.>Vendor dimension evaluation weight modifier +.>Based on the interaction between the vendor supply product quality level and the vendor company scale factor and function, a functional relationship is formed:

；

through the interaction relation between the quality level of the products supplied by the suppliers and the delivery accuracy of the products supplied by the suppliers, the dimension evaluation of the suppliers is carried out under the condition of ensuring the accuracy of dimension evaluation data, the supply stability coefficient of the products supplied by the suppliers and the total evaluation contribution value of the suppliers are generated, and the supply dimension evaluation weight correction quantity is utilized to reduce the data redundancy under the condition of ensuring the accuracy of the data, thereby saving the calculation force, leading the calculation to be converged quickly and leading the calculation to be carried out under the condition of ensuring the accuracy of the dataAdjusting the supplier dimension evaluation dimension to generate the supplier dimension evaluation weight more accurately>The dimension evaluation of suppliers is improvedAccuracy and reliability of the price. Meanwhile, parameters such as the after-sale service quality coefficient of the provider product, the supply stability coefficient of the provider product and the like in the formula can be adjusted according to actual conditions, so that the method is suitable for different dimension evaluation scenes, and the applicability and the flexibility of the algorithm are improved.

Preferably, step S24 comprises the steps of:

step S241: performing dimension weight division processing on the primary dimension evaluation data and the secondary dimension evaluation data of the suppliers so as to generate a primary dimension weight value and a secondary dimension weight value;

step S242: respectively carrying out dimension data weighting processing on the primary dimension weight value and the secondary dimension weight value by utilizing an entropy weight method to generate a primary dimension weight value and a secondary dimension weight value;

step S243: performing weight combination processing on the primary dimension weighted value and the secondary dimension weighted value by using a weighted average division method to generate comprehensive evaluation data; and performing evaluation result sorting processing on the comprehensive evaluation data so as to generate supplier dimension evaluation data.

According to the method, the primary dimension evaluation data and the secondary dimension evaluation data of the suppliers are subjected to dimension weight division processing, so that a primary dimension weight value and a secondary dimension weight value are generated, which dimensions have more important influence on an evaluation result can be determined, the evaluation accuracy is improved, the suppliers are evaluated more objectively and scientifically, powerful support is provided for purchasing decisions, the importance of different dimensions in evaluation indexes is known, and therefore an evaluation index system is further optimized, and the evaluation effect is improved; the main dimension weight value and the secondary dimension weight value are respectively subjected to dimension data weighting processing by utilizing an entropy weight method, so that the main dimension weight value and the secondary dimension weight value are generated, comprehensive evaluation of suppliers can be evaluated more scientifically, more accurate data support is provided for purchasing decisions, the scientificity and the accuracy of purchasing decisions are improved, the influence of subjective factors is reduced, the weight of evaluation indexes is optimized, and the suppliers can be compared and screened conveniently; performing weight combination processing on the primary dimension weighted value and the secondary dimension weighted value by using a weighted average division method to generate comprehensive evaluation data; and (3) carrying out evaluation result sequencing treatment on the comprehensive evaluation data so as to generate supplier dimension evaluation data, so that the scientificity and accuracy of purchasing decisions can be improved, the advantages and disadvantages of suppliers are highlighted, evaluation indexes are optimized, and the grade and risk grade of the suppliers are defined.

Preferably, step S3 comprises the steps of:

step S31: performing data exploratory analysis processing on the provider dimension evaluation data to generate supply exploratory data; carrying out data feature selection processing on the supply exploration data to generate supply feature data;

step S32: carrying out supply data division processing on the supply characteristic data by using a hierarchical sampling separation method to generate supply training set data and supply test set data;

step S33: carrying out data modeling processing on the provider dimension evaluation data by using a Logistic regression algorithm to generate a pre-supply model; model training processing is carried out on the pre-supply model by utilizing supply training set data, and a supply model is generated;

step S34: performing model fitting test processing on the supply model so as to generate a standard supply model;

step S35: carrying out probability value prediction processing on the provider dimension evaluation data through a standard supply model to generate a supply result probability value; and carrying out probability value score calibration processing on the probability value of the supply result to generate a supply evaluation score value.

According to the invention, the supplier dimension evaluation data is subjected to data exploratory analysis processing to generate supply exploration data, abnormal values and outliers can be found, so that the data quality problem is helped to be identified, the accuracy and the reliability of the data are improved, the association and the trend in the data are found, the supply exploration data is subjected to data feature selection processing to generate supply feature data, the accuracy and the reliability of the data can be improved, and the accuracy and the reliability of an evaluation model are improved; the supply characteristic data is divided by using a hierarchical sampling separation method to generate supply training set data and supply test set data, so that the supply data can be used for training and testing a model at the same time, and the accuracy and reliability of an evaluation result are improved by fully utilizing the data; the Logistic regression algorithm is utilized to carry out data modeling processing on the supplier dimension evaluation data to generate a pre-supply model, so that the accuracy and reliability of model establishment can be improved, the pre-supply model is utilized to carry out model training processing on the supply training set data to generate a supply model, the quick and accurate analysis of the relationship between the data can be facilitated, and the decision efficiency can be improved; performing model fitting test processing on the supply model to generate a standard supply model, ensuring the accuracy and reliability of the model, judging whether the model has over-fitting or under-fitting problems, and improving the generalization and adaptability of the model; the probability value prediction processing is carried out on the provider dimension evaluation data through the standard supply model to generate a supply result probability value, the provider performance can be finely analyzed to find out the advantages and disadvantages of the provider, the efficiency of supply chain management is improved, the probability value score calibration processing is carried out on the supply result probability value to generate a supply evaluation score value, the error and uncertainty of the evaluation result can be eliminated, the evaluation result is more accurate and objective, the reliability and stability of the evaluation result are improved, and a standard system of the provider performance evaluation is established, so that the performance evaluation index and the corresponding score range of the provider are established, the automatic provider evaluation and update are realized, the manual operation and error are reduced, and the decision efficiency and accuracy are improved.

Preferably, step S34 includes the steps of:

step S341: performing data sparsification processing on the supply test set data to generate supply sparsification data;

step S342: performing data fitting processing on the supply sparse data by using a normal distribution function to generate a supply fitting curve; performing curve coincidence processing on the supplied fitting curve and a preset standard fitting curve, marking the completely coincident region as an overfitting region, judging the fitting distance of the region which is not coincident by using a fitting distance judgment formula, generating an underfitting region when the fitting distance is larger than the fitting distance, and generating a fitting normal region when the fitting distance is smaller than the fitting distance;

step S343: and carrying out data regularization treatment on the over-fitting region until the fitting normal region is regenerated, and carrying out regularization degree reduction treatment on the under-fitting region until the fitting normal region is regenerated, so as to obtain the standard supply model.

According to the invention, the data sparsification processing is carried out on the data of the supply test set to generate the supply sparsification data, and the denser part in the data set can be filtered and reserved, so that the data redundancy is reduced, and the calculation efficiency is improved; carrying out data fitting processing on the supply sparse data by utilizing a normal distribution function to generate a supply fitting curve, researching the distribution rule and probability distribution situation of the supply sparse data, so that the characteristics and rule of data distribution can be better understood, the trend and change of the supply data are predicted, a more accurate and effective evaluation index is constructed, the scientificity and statistical reliability of an evaluation standard and a method are improved, the supply fitting curve and a preset standard fitting curve are subjected to curve overlapping processing, a completely overlapped area is marked as an overfitting area, fitting distance judgment is carried out on the area which is not overlapped by utilizing a fitting distance judgment formula, an underfitting area is generated when the fitting distance is larger than the fitting distance, a fitting normal area is generated when the fitting distance is smaller than the fitting distance, and the fitting effect can be evaluated, the model parameter selection is optimized, the model visualization effect is improved and an evaluation standard system is established; and carrying out data regularization treatment on the over-fitting region until the fitting normal region is regenerated, carrying out regularization degree reduction treatment on the under-fitting region until the fitting normal region is regenerated, thereby obtaining a standard supply model, reducing the risk of over-fitting and under-fitting, and further improving the generalization capability and prediction precision of the model.

Preferably, the fitting distance judgment formula in step S342 is specifically as follows:

；

in the method, in the process of the invention,represents the final distance of the fit, +.>Represents the actual distance of the fitted curve, +.>Represents the predicted distance of the fitting curve, +.>Representing the actual distance weight of the fitted curve, +.>Representing the predicted distance weight of the fitted curve, +.>Represents the fitting curve deviation adjustment coefficient,/->Represents the coefficient of contraction of the fitted curve, +.>Represents the number of samples in the region of the fitted curve, +.>Representation and->Regional sample coefficients with samples nearest to each other, +.>Representing the index of the sample of the fitted curve +.>And the fitted curve distance judgment abnormal value is represented.

The invention constructs a fitting distance judging formula which fully considers the actual distance of a fitting curveFitting curve predicted distance +.>Fitting curveLine actual distance weight +.>Fitting curve prediction distance weight +.>Fitting curve deviation adjustment coefficient ∈ ->Fitting curve shrinkage factor->Number of samples of fitted curve region ∈ ->And->Regional sample coefficients with nearest sample distance +.>Fitting Curve sample index->Judging abnormal value by fitting curve distance>According to the interaction between the actual distance of the fitting curve and the predicted distance and the function of the fitting curve, the functional relation is formed:

；

Through the interaction relation of the actual distance weight of the fitting curve and the predicted distance weight of the fitting curve, the fitting distance measurement is carried out under the condition of ensuring the accuracy of the fitting curve distance, the fitting curve shrinkage coefficient and the number of samples in the fitting curve area are generated, the fitting curve sample index is utilized, the data redundancy is reduced under the condition of ensuring the accuracy of the data, the calculation force is saved, the calculation is enabled to achieve rapid convergence, and the judgment is carried out through the fitting curve distanceOutlier valueThe distance of the fitting curve is adjusted, and the fitting final distance is generated more accurately>The accuracy and the reliability of the distance measurement of the fitting curve are improved. Meanwhile, parameters such as the number of the fitted curve area samples, the fitted curve deviation adjustment coefficient and the like in the formula can be adjusted according to actual conditions, so that the method is suitable for different fitted curve scenes, and the applicability and the flexibility of the algorithm are improved.

Preferably, step S4 comprises the steps of:

step S41: dividing the supplier grade according to the supply demand and the evaluation standard, wherein the supplier grade is divided into a high-quality supplier, a qualified supplier, a trial supplier, a standby supplier and a blacklist supplier;

step S42: carrying out value interval division processing on the supply evaluation score value according to the grade of the supplier to generate a supplier evaluation score interval;

Step S43: and carrying out score comparison processing on the supply evaluation score value by using the provider evaluation score interval so as to generate a provider score rating.

The invention can realize the fine management of suppliers by dividing the grades of the suppliers according to the supply requirements and the evaluation standards. Different management strategies can be adopted by suppliers with different grades to ensure the maintenance of high-quality suppliers and the guidance and optimization of qualified suppliers, improve the stability of a supply chain, optimize the flow and efficiency of the supply chain and improve the credibility of the suppliers; the supply evaluation score value is subjected to value interval division according to the provider grade, so that a provider evaluation score interval is generated, interference of subjective factors can be reduced, and fairness and objectivity of evaluation are improved; and the score comparison processing is carried out on the supplied evaluation score value by using the supplier evaluation score interval, so that the supplier score rating is generated, the supplier rating can be more intuitively and rapidly realized, and meanwhile, the grouping management of suppliers is facilitated.

Preferably, step S5 comprises the steps of:

step S51: performing decision training on the score ratings of the suppliers by utilizing a decision tree model; generating a first level decision when the provider score is rated as a premium provider; generating a second level decision when the vendor score is rated as a qualified vendor; generating a third level decision when the vendor score is rated as the trial vendor; generating a fourth level decision when the provider score is rated as a spare provider; generating a fifth level decision when the provider score is rated as a blacklisted provider;

Step S52: when the decision tree model receives the first level decision, performing behavior decision analysis processing on the first level decision to generate an excellent supplier evaluation scheme; when the decision tree receives the second level decision, performing behavior decision analysis processing on the second level decision to generate a qualified provider evaluation scheme; when the decision tree receives the third-level decision, performing behavior decision analysis processing on the third-level decision to generate a trial provider evaluation scheme; when the decision tree receives the fourth-level decision, carrying out decision analysis processing on the fourth-level decision to generate a standby provider evaluation scheme; and when the decision tree receives the fifth-level decision, carrying out decision analysis processing on the fifth-level decision to generate a blacklist provider evaluation scheme.

According to the invention, by utilizing the decision tree model to carry out decision training on the score rating of the provider, a plurality of factors can be comprehensively considered to obtain the score rating of the provider, and corresponding decisions are generated according to the rating result. Therefore, the grading of the suppliers is more scientific and accurate, the influence of individual factors or subjective factors on the grading result is avoided, the objectivity and accuracy of the evaluation are improved, the efficiency and accuracy of the management of the suppliers are improved, the objectivity and transparency are enhanced, and the creditability and the credibility of the management are improved; by using the decision tree model to generate different evaluation schemes for different level decisions, the quality of suppliers can be improved, the quality of qualified suppliers is prevented from being reduced, an improvement space is provided for trial suppliers, the management of blacklist suppliers is enhanced, and the scientificity and transparency of the management are enhanced.

The invention checks the data by acquiring the supplier supply data, because the data types provided by the suppliers can be inconsistent, the data volume is huge and the repeated or missing data possibly exists, the supplier supply data is required to be processed, firstly, the supply data is preprocessed to reduce adverse effects caused by useless, redundant and abnormal data, and the supplier supply data is divided into primary dimension evaluation data and secondary dimension evaluation data by dimension evaluation division so as to grade the supplier dimension, the primary dimension evaluation data and the secondary dimension evaluation data are weighted by a supplier dimension evaluation weight formula, the evaluation result can be quantized, thereby realizing the visualization of the bottom data, the supplier dimension evaluation data is subjected to data exploratory analysis processing, the data is divided into supply training set data and supply test set data, carrying out data modeling processing and training processing on provider dimension evaluation data by using a Logistic regression algorithm to generate a standard supply model, which can help to quickly and accurately analyze the relationship between the data, improve decision efficiency, and because the problem of over-fitting or under-fitting possibly exists between the models, the model needs to be established and corrected by carrying out distance judgment on the fitting curve by using a fitting distance judgment formula, carrying out data regularization processing on the over-fitting region until a fitting normal region is regenerated, carrying out regularization degree relieving processing on the under-fitting region until the fitting normal region is regenerated, thereby obtaining the standard supply model, improving model generalization capability and prediction precision, dividing provider grades according to supply requirements and evaluation standards, carrying out numerical interval division processing on supply evaluation score values according to the provider grades, the method comprises the steps of generating a provider evaluation score interval, carrying out score comparison processing on a provider evaluation score value by using the provider evaluation score interval, so that provider score rating is generated, provider rating can be more intuitively and rapidly achieved, provider grouping management is facilitated, decision training and analysis processing are carried out on the provider score rating by using a decision tree model, different level schemes are generated, objectivity and accuracy of evaluation are improved, efficiency and accuracy of provider management are improved, objective fairness and transparency are improved, and accordingly public belief and credibility of management are improved, and quality of providers can be improved. Therefore, the artificial intelligence provider data model construction method provided by the invention carries out provider grade grading on provider data, and establishes a provider evaluation model by using a Logistic regression algorithm so as to realize the decision support for the admission of provider projects, reduce the labor cost and improve the model precision.

Drawings

FIG. 1 is a schematic flow chart of steps of a method for constructing a vendor supply data model based on artificial intelligence;

FIG. 2 is a detailed flowchart illustrating the implementation of step S1 in FIG. 1;

FIG. 3 is a detailed flowchart illustrating the implementation of step S2 in FIG. 1;

FIG. 4 is a flowchart illustrating the detailed implementation of step S24 in FIG. 1;

FIG. 5 is a flowchart illustrating the detailed implementation of step S3 in FIG. 1;

the achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.

Detailed Description

The following is a clear and complete description of the technical method of the present patent in conjunction with the accompanying drawings, and it is evident that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, are intended to fall within the scope of the present invention.

Furthermore, the drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus a repetitive description thereof will be omitted. Some of the block diagrams shown in the figures are functional entities and do not necessarily correspond to physically or logically separate entities. The functional entities may be implemented in software or in one or more hardware modules or integrated circuits or in different networks and/or processor methods and/or microcontroller methods.

It will be understood that, although the terms "first," "second," etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another element. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element, without departing from the scope of example embodiments. The term "and/or" as used herein includes any and all combinations of one or more of the associated listed items.

To achieve the above object, referring to fig. 1 to 5, a method for constructing a provider supply data model based on artificial intelligence, the method comprising the steps of:

According to the invention, the supplier supply data is obtained and subjected to data preprocessing, so that noise and errors in the data can be reduced, the accuracy and reliability of the data are improved, the data processing cost is reduced, the data processing time is shortened, the data consistency and comparability are improved, and standard supply data are generated; the standard supply data is subjected to evaluation division processing according to the preset evaluation dimension, so that supplier dimension evaluation data is generated, the comprehensiveness and accuracy of evaluation can be improved, self-improvement of suppliers is promoted, and supplier selection and management strategies are optimized; the Logistic regression algorithm is used for carrying out Logistic regression processing on the provider dimension evaluation data, a supply evaluation model can be established, and the performance of the provider is accurately measured, so that the accuracy and the reliability of the provider evaluation result are improved, the evaluation process is standardized, the influence of artificial subjective factors on the evaluation result is reduced, the possibility of difference of the evaluation result is reduced, the accuracy and the fairness of the provider evaluation result are improved, and the continuous improvement and improvement of the provider in the aspects of quality, cost, delivery period and the like are promoted; the score rating mapping processing is carried out on the supply evaluation score value through the preset supplier grade, so that the supplier score rating is generated, the evaluation result is more visual and easy to understand, the risk is reduced, the stability of a supply chain is improved, the competition consciousness and innovation enthusiasm of suppliers are stimulated, the performance and the service quality are further improved, the enterprise is facilitated to optimize purchasing decisions, and the purchasing efficiency and the supply chain management effect are improved; and grading decision processing is carried out on the score ratings of the suppliers by utilizing a decision tree model, so that a decision result of the suppliers is generated, and classification management can be carried out on the grading data of the suppliers rapidly, so that the decision efficiency is improved, risks in purchasing decision and supply chain management are reduced, all parties in a supply chain are better managed, the supply chain management effect is improved, purchasing flows are standardized, and the management cost is reduced. Therefore, the artificial intelligence provider data model construction method provided by the invention carries out provider grade grading on provider data, and establishes a provider evaluation model by using a Logistic regression algorithm so as to realize the decision support for the admission of provider projects, reduce the labor cost and improve the model precision.

In the embodiment of the present invention, as described with reference to fig. 1, a schematic flow chart of steps of a method for constructing an artificial intelligence-based vendor supply data model of the present invention is provided, and in this example, the method for constructing an artificial intelligence-based vendor supply data model includes the following steps:

in the embodiment of the invention, the data preprocessing is performed on the supplier supply data by acquiring the supplier supply data, wherein the data preprocessing comprises the following steps: missing value filling processing, outlier detection processing, and data denoising, thereby generating standard supply data.

in the embodiment of the invention, the provider evaluation dimension is preset according to the requirement, wherein the provider evaluation dimension comprises the product quality level, the product response speed, the product delivery accuracy, the product cost, the after-sale service and the like. And carrying out evaluation analysis on the suppliers according to the dimensions, carrying out corresponding weighting processing on the scores of different evaluation standards, and calculating the overall evaluation score of the suppliers so as to generate the dimension evaluation data of the suppliers.

in the embodiment of the invention, the standard supply model is generated by determining the evaluation dimension and the evaluation index, collecting the training data, splitting the training data into the training data and the test data, performing model training by using a Logistic regression algorithm, and evaluating the model by using the test data. And importing the supplier dimension evaluation data into a standard supply model, performing supply prediction processing by using the model to generate a supplier score, performing normalization processing on the supplier score, and sequencing the suppliers according to the score to generate a supply evaluation score value.

in the embodiment of the invention, the supplier class is divided into five classes by establishing a supplier class system, which are respectively: quality suppliers, qualified suppliers, trial suppliers, spare suppliers, and blacklist suppliers. And mapping the score of the provider to a specific grade by adopting a linear or nonlinear mapping relation, and converting the score of the provider to a specific grade according to the score mapping relation, so that the grade of the score of the provider can be generated.

In the embodiment of the invention, according to the provider score rating data, a decision tree algorithm such as ID3, C4.5 or CART is utilized to divide the data set into a training set and a testing set randomly, and the training set is used for training the decision tree model. In the training process, pruning can be carried out on the decision tree to avoid the situation of fitting, accuracy and performance of the model are verified through a test set, the model is adjusted and optimized to form a decision tree model, the score of the supplier is used as input of the decision tree, and the score of the supplier is gradually classified along branches of the tree to obtain a decision result of the supplier.

Preferably, step S1 comprises the steps of:

As an example of the present invention, referring to fig. 2, a detailed implementation step flow diagram of step S1 in fig. 1 is shown, where step S1 includes:

in the embodiment of the invention, the data missing value filling process is performed on the supplier supplied data, wherein the missing value filling process comprises the following steps: firstly, checking and analyzing the data supplied by the supplier to determine which data have missing values; the method for filling the missing values of the supplier supply data of different data types comprises the steps of carrying out batch processing filling operation on the supplier supply data by using average filling, median filling, nearest neighbor filling, difference filling, regression filling and other methods, and after filling, checking and verifying the filling data to ensure that the filling result meets the expected requirement and generating the supply filling data.

in the embodiment of the invention, the outlier detection method, such as a standard method, a box diagram method, a K nearest neighbor method, a DBSCAN clustering method and the like, is utilized to detect outliers of the supplied filling data, if the outliers of partial data are found in the outlier detection process, log-scale logarithmic transformation is carried out on the outlier data, so that the data are ensured to meet normal distribution or approximate normal distribution, and outliers are eliminated, thereby generating the supplied outlier data.

In the embodiment of the invention, the supply outlier data is subjected to denoising processing, wherein the data denoising includes using methods such as mean filtering, median filtering and Gaussian filtering, so as to obtain supply denoising data, carrying out data normalization processing on the supply denoising data by using quantile normalization, converting the data into a numerical value of a designated quantile, and generating standard supply data.

Preferably, step S2 comprises the steps of:

As an example of the present invention, referring to fig. 3, a detailed implementation step flow diagram of step S2 in fig. 1 is shown, where step S2 includes:

in the embodiment of the invention, the supply weight calculation processing is carried out on the supplier rating dimension data according to the supplier rating weight formula by acquiring the supplier rating dimension data, the weight values of all suppliers in different dimensions are calculated according to the supplier rating and the weight coefficient, and the total weight is obtained by adding the weight values, so that the supply rating weight coefficient value is generated.

in the embodiment of the invention, the standard supply data is subjected to weight comparison processing according to the supply weight coefficient value, wherein the weight comparison processing step comprises the following steps: the performance of the provider in each dimension is scored. For example, in the supplier stability dimension, the stability of the suppliers may be evaluated based on the on-time delivery rate, delivery time standard deviation, and the like of the suppliers, and the score of each supplier in each dimension may be weighted based on the weight coefficient of the evaluation dimension. For example, if a certain provider has a score of 80 in a "provider stability dimension" and a weight coefficient of 0.3, the weighted score of the dimension is 80×0.3=24, and the weighted scores of all the evaluation dimensions of each provider are summed to obtain a total weighted score of the provider, so as to generate the supply weight comparison data.

in the embodiment of the invention, the supply weight comparison data and the preset standard supply weight comparison data are subjected to weight comparison processing, and when the supply weight comparison data is larger than the standard supply weight, the supply weight comparison data are divided into main evaluation dimensions of suppliers so as to generate main dimension evaluation data of the suppliers; dividing the supply weight comparison data into a supplier secondary evaluation dimension when the supply weight comparison data is smaller than the standard supply weight, so as to generate supplier secondary dimension evaluation data, wherein the supplier primary dimension evaluation data comprises information such as scores, ranks, score distribution conditions and the like of suppliers in the primary evaluation dimension; the provider secondary dimension rating data comprises information such as scores, ranks, score distribution conditions and the like of the provider in the secondary rating dimension.

In the embodiment of the invention, the scores of the primary evaluation dimension and the secondary evaluation dimension of each provider are ordered, and the ranking of each provider in each evaluation dimension is determined. And comprehensively evaluating the performances of different suppliers according to the ranking condition of each evaluation dimension, and generating the evaluation data of the suppliers dimension. These data may reflect the behavior of different suppliers in the various evaluation dimensions, generating supplier dimension evaluation data.

；

in the method, in the process of the invention,representing vendor dimension rating weight, +.>Indicating the level of quality of the product supplied by the supplier, < >>Representing the response speed of the supplier to supply the product, +.>Indicating the accuracy of delivery of the product supplied by the supplier, < >>Representing the cost of the product supplied by the supplier, < >>Representing the minimum consideration condition number of the supplier evaluation, < ->Representing the maximum consideration condition number of the supplier evaluation, < ->Representing the supplier company size factor,/-, for>Representing the after-sales quality of service coefficient of the provider product, < > >Representing the supplier product supply stability factor, +.>Representing the total score contribution value of the supplier supply, < ->Representing the vendor dimension rating weight modifier.

The invention constructs a supplier dimension evaluation weight formula which fully considers the quality level of the products supplied by suppliersResponse speed of vendor supplied products>Delivery accuracy of products supplied by suppliers->Cost of product supplied by suppliers->Supplier evaluation minimum consideration condition number +.>Supplier evaluation maximum consideration condition number +.>Vendor company Scale factor->After-sales service quality factor of vendor product>Stability factor of the supplier's product supply>Total score contribution value supplied by supplier +.>Vendor dimension evaluation weight modifier +.>Based on the interaction between the vendor supply product quality level and the vendor company scale factor and function, a functional relationship is formed:

；

through the interaction relation between the quality level of the products supplied by the suppliers and the delivery accuracy of the products supplied by the suppliers, the dimension evaluation of the suppliers is carried out under the condition of ensuring the accuracy of dimension evaluation data, the supply stability coefficient of the products supplied by the suppliers and the total evaluation contribution value of the suppliers are generated, and the supply dimension evaluation weight correction quantity is utilized to reduce the data redundancy under the condition of ensuring the accuracy of the data, thereby saving the calculation force, leading the calculation to be converged quickly and leading the calculation to be carried out under the condition of ensuring the accuracy of the data Adjusting the supplier dimension evaluation dimension to generate the supplier dimension evaluation weight more accurately>Accuracy and accessibility of supplier dimension evaluation are improvedReliability. Meanwhile, parameters such as the after-sale service quality coefficient of the provider product, the supply stability coefficient of the provider product and the like in the formula can be adjusted according to actual conditions, so that the method is suitable for different dimension evaluation scenes, and the applicability and the flexibility of the algorithm are improved.

Preferably, step S24 comprises the steps of:

As an example of the present invention, referring to fig. 4, a detailed implementation step flow diagram of step S24 in fig. 3 is shown, where step S24 includes:

step S241: performing data exploratory analysis processing on the provider dimension evaluation data to generate supply exploratory data; carrying out data feature selection processing on the supply exploration data to generate supply feature data;

in the embodiment of the invention, the primary dimension and the secondary dimension evaluation data of the suppliers are subjected to dimension weight division processing, for example, for supply chain management of a manufacturing enterprise, the delivery stability, the price rationality, the product quality and the like of the suppliers may be important evaluation indexes, and the indexes can be divided into primary dimensions, while other indexes are divided into secondary dimensions, so that a primary dimension weight value and a secondary dimension weight value are generated.

in the embodiment of the invention, the main dimension weight value and the secondary dimension weight value are respectively subjected to dimension data weighting processing by utilizing an entropy weight method, the entropy value of each dimension is calculated, then the weight coefficient of each dimension is calculated, for the main dimension, the weight coefficient of the main dimension is multiplied by the score of the evaluation index corresponding to the weight coefficient to calculate the aggregation score of the main dimension, the main dimension weight value is generated, and for the secondary dimension, the weight coefficient of the secondary dimension is multiplied by the score of the evaluation index corresponding to the weight coefficient to calculate the aggregation score of the secondary dimension, and the secondary dimension weight value is generated.

In the embodiment of the invention, a weighted average division method is utilized to carry out weight combination processing on the primary dimension weighted value and the secondary dimension weighted value, comprehensive evaluation data is generated, the comprehensive evaluation data of all suppliers are sequenced by the comprehensive evaluation data, and the supplier dimension evaluation data is obtained to be divided according to the ranking. For example, vendor scores are divided into premium vendors at 90-100, qualified vendors at 80-89, trial vendors at 70-79, spare vendors at 60-69, and blacklist vendors below 60, thereby generating vendor dimension rating data.

Preferably, step S3 comprises the steps of:

As an example of the present invention, referring to fig. 5, a detailed implementation step flow diagram of step S3 in fig. 1 is shown, where step S3 includes:

in the embodiment of the invention, the data exploratory analysis is carried out on the provider dimension evaluation data, the information such as distribution, correlation, abnormal values and the like of graphic display data such as a histogram, a scatter diagram, a box diagram and the like is utilized to generate the supply exploratory data, the data feature selection processing is carried out on the supply exploratory data, and the text features in the supply data are extracted, so that the supply feature data are generated.

in the embodiment of the invention, the supply characteristic data is layered according to some characteristics of a provider, such as a product type, a geographic position and the like, the sampling proportion is designated according to the supply characteristic data sample, the sampling proportion is designated according to the size and the overall size of the supply characteristic data sample, the samples are randomly extracted with corresponding probability for each layer, the samples randomly extracted in each layer are put together to form a supply data set, for example, 70% of data is used as supply training set data according to the proportion of 7:3, and the rest 30% of data is used as supply test set data.

in the embodiment of the invention, the data modeling processing is performed on the provider dimension evaluation data by using a Logistic regression algorithm, wherein the data modeling processing comprises the following steps: performing number cleaning and normalization processing on the provider dimension evaluation data; and carrying out parameter estimation on the model by adopting a maximum likelihood estimation method to obtain a pre-supply model, evaluating the pre-supply model by using pre-supply test set data, and evaluating indexes such as accuracy, precision, recall rate and the like of the model so as to generate the supply model.

in the embodiment of the invention, a two-dimensional data curve and a polynomial fitting function in a supply model are subjected to curve comparison, a coincident region and a non-coincident region are intercepted, wherein the coincident region is an over-fitting region, the two-dimensional data curve and the polynomial fitting function curve in the non-coincident region are subjected to vertical axis distance comparison, whether the two-dimensional data curve is an under-fitting region or not is judged, and regularized adjustment processing is carried out on the over-fitting region and the under-fitting region, so that a standard supply model is generated.

In the embodiment of the invention, probability value prediction is performed on the provider dimension evaluation data by using a standard supply model, and a probability value with a result of 0-1 is generated and is expressed as a supply result probability value of an excellent provider, wherein the probability value is calculated as follows:

；

wherein the method comprises the steps ofRepresented as a probability value offset, where the offset is in the interval 0,1]On (I)>Expressed as an evaluation base score->Expressed as an evaluation factor>Expressed as the ratio of the probability of default to the probability of normal, < >>Abbreviated as Points to Double the Odds, the odds doubles, increasing the score. For example, if the odds increase from 100:1 to 200:1, the score will change by how much, a common default for PDO is 20, as it will produce a credit rating range that people tend to like.Expressed as calibration score +.>Expressed as customer breach probability. The above procedure calculates the total score of a single sample, provides the sub-score corresponding to each group in each variable, and the calculation method of the sub-score is as follows: / >

；

Wherein,,expressed as sub-score, ">Expressed as number of scores>Expressed as score intercept->Values expressed as categories in discrete variables, +.>Expressed as sub-score variable coefficients, by the above formula we can translate the probability value of 0-1 into an arbitrary range of values, thereby generating a supply score value, e.g., 0-100 points.

Preferably, step S34 includes the steps of:

In the embodiment of the invention, the data of the supply test set is subjected to data sparsification processing by using a sparse matrix method, the data is presented in a sparse matrix form, the supply sparsification data is generated, and the normal distribution function is used for fitting the supply sparsification data, so that a fitting curve of the supply data is obtained. And simultaneously, carrying out curve coincidence processing on the supply fitting curve and a preset standard fitting curve, wherein the preset standard fitting curve is a standard fitting curve generated by utilizing a polynomial fitting function, obtaining a coincidence curve through curve coincidence processing, intercepting a coincident region and a non-coincident region, marking the coincident region as an overfitting region, and carrying out vertical axis distance detection and judgment on the non-coincident region to obtain an underfitting region. Regularization treatment is carried out on the over-fitting area until a fitting normal area is regenerated; and carrying out regularization degree reducing treatment on the under-fitting region until the fitting normal region is regenerated, so as to obtain a standard supply model.

；

in the method, in the process of the invention,represents the final distance of the fit, +.>Represents the actual distance of the fitted curve, +. >Represents the predicted distance of the fitting curve, +.>Representing the actual distance weight of the fitted curve, +.>Representing the predicted distance weight of the fitted curve, +.>Represents the fitting curve deviation adjustment coefficient,/->Represents the coefficient of contraction of the fitted curve, +.>Represents the number of samples in the region of the fitted curve, +.>Representation and->Regional sample coefficients with samples nearest to each other, +.>Representing the index of the sample of the fitted curve +.>And the fitted curve distance judgment abnormal value is represented.

The invention constructs a fitting distance judging formula which fully considers the actual distance of a fitting curveFitting curve predicted distance +.>Fitting curve actual distance weight +.>Fitting curve prediction distance weight +.>Fitting curve deviation adjustment coefficient ∈ ->Quasi-Coefficient of contraction of the resultant curve->Number of samples of fitted curve region ∈ ->And->Regional sample coefficients with nearest sample distance +.>Fitting Curve sample index->Judging abnormal value by fitting curve distance>According to the interaction between the actual distance of the fitting curve and the predicted distance and the function of the fitting curve, the functional relation is formed:

；

through the interaction relation of the actual distance weight of the fitting curve and the predicted distance weight of the fitting curve, the fitting distance measurement is carried out under the condition of ensuring the accuracy of the fitting curve distance, the fitting curve shrinkage coefficient and the number of samples in the fitting curve area are generated, the fitting curve sample index is utilized, the data redundancy is reduced under the condition of ensuring the accuracy of the data, the calculation force is saved, the calculation is enabled to achieve rapid convergence, and the abnormal value is judged through the fitting curve distance The distance of the fitting curve is adjusted, and the fitting final distance is generated more accurately>The accuracy and the reliability of the distance measurement of the fitting curve are improved. At the same time, the number of the samples of the fitting curve area and the fitting curve in the formulaParameters such as the line deviation adjusting coefficient and the like can be adjusted according to actual conditions, so that the method is suitable for different fitting curve scenes, and the applicability and the flexibility of the algorithm are improved.

Preferably, step S4 comprises the steps of:

In the embodiment of the invention, suppliers are classified into high-quality suppliers, qualified suppliers, trial suppliers and blacklist supplier grades according to supply requirements and evaluation standards. Specific criteria may be assessed and classified based on performance assessment metrics, supply capacity, compliance level, delivery time, and price. Mapping the evaluation score obtained by the supplier to the corresponding supplier grade, and dividing the evaluation score interval of each grade, for example: the high-quality suppliers have a score of more than 90, the qualified suppliers have a score of 80-89, the trial suppliers have a score of 70-79, the standby suppliers have a score of 60-69, and the blacklist suppliers have a score of less than 60. For each evaluation index, the score is mapped to a corresponding grading grade according to the score range and the score interval, for example, a certain supplier is classified into a high-quality supplier when the score is 85 in the delivery time index, the grading obtained by all the evaluation indexes is comprehensively considered, and the final score grade of the supplier is calculated according to comprehensive evaluation methods such as a score weighted average method and the like to obtain the score grade of the supplier.

Preferably, step S5 comprises the steps of:

In the embodiment of the invention, decision training is carried out on different provider score grades by utilizing a decision tree model, and when the provider score grades are high-quality providers, a first grade decision is generated; generating a second level decision when the vendor score is rated as a qualified vendor; generating a third level decision when the vendor score is rated as the trial vendor; generating a fourth level decision when the provider score is rated as a spare provider; generating a fifth level decision when the provider score is rated as a blacklisted provider; for example: suppliers are classified as premium suppliers, and corresponding first level decision making schemes, such as quick approval of orders, preferential payments, etc., are generated. Suppliers are classified as qualified suppliers, and corresponding second-level decision schemes, such as post-order approval, normal payment, etc., are generated. Classifying suppliers as trial suppliers, generating corresponding third-level decision schemes, such as gradually increasing order quantity, prolonging payment time and the like; suppliers are classified as spare suppliers, and corresponding fourth level decision schemes are generated, such as periodic tracking of performance and supply capacity, expedited process auditing, and the like. Classifying the suppliers into blacklist suppliers, generating a corresponding fifth-level decision scheme, and refusing the blacklist suppliers to enter the project supply link.

The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.

The foregoing is only a specific embodiment of the invention to enable those skilled in the art to understand or practice the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. The provider supply data model construction method based on artificial intelligence is characterized by comprising the following steps:

2. The artificial intelligence based vendor provisioning data model construction method of claim 1, wherein step S1 comprises the steps of:

3. The artificial intelligence based vendor provisioning data model construction method of claim 2, wherein step S2 comprises the steps of:

4. The artificial intelligence based vendor supply data model building method of claim 3, wherein the vendor dimension evaluation weight formula in step S21 is as follows:

；

5. The artificial intelligence based vendor provisioning data model construction method of claim 3, wherein step S24 comprises the steps of:

6. The artificial intelligence based vendor provisioning data model construction method of claim 3, wherein step S3 comprises the steps of:

7. The artificial intelligence based vendor provisioning data model construction method of claim 6, wherein step S34 comprises the steps of:

8. The artificial intelligence based vendor supply data model building method according to claim 7, wherein the fitting distance judgment formula in step S342 is as follows:

；

9. The artificial intelligence based vendor provisioning data model construction method of claim 6, wherein step S4 comprises the steps of:

10. The artificial intelligence based vendor provisioning data model construction method of claim 9, wherein step S5 comprises the steps of: