The Research And Application Of The Assessment System Of Suppliers Based On The SVM And Decision Tree Theory

Reasonable evaluation and supplier selection for the modern enterprise is the key,which directly influents production activities,the quality of product,production cycle,corporate profits and so on.Along with economic development and scientific technological progress,enterprises continue to expand access to information channel with increasing the number of candidate suppliers.However,supplier evaluation becomes more complicated.On the one hand,with the increase of amount of information,the uncertainty and incompleteness of supplier informati on is more and more obvious.It needs to adopt the way of group decision by analyzing the information of the supplier decision-making,increase the objectivity of decision results and robustness.On the other hand,how to use data mining method to extract the related supplier evaluation and select history program data contain ing in the relevant information,to guide enterprises to carry out reasonable and effective decisions,is also a need to solve.Based on the weakness of current research,the main innov ation results are as follows:(1)In view of the uncertainty and incompleteness decision-making information,the evaluation method is proposed based on two-dimensional hybrid evaluation information.This method at different time points for each supplier to maximize deviation analysis to calculate the weights of decision makers.This method restores the subjectivity of the assessment information and weights calculation process objective.On this basis,draw lessons from normal random variable distribution method to deal effectively with two-dimensional hybrid assessment information,better retain its uncertainty.Last degree integration and growth characteristics of order of the supplier for final evaluation and selection,form a complete set of processing incomplete information supplier evaluation method.(2)According to supplier select problem,the theory of pattern recognition is considered.Consider the specific needs of the characteristics of supplier selection,classification characteristic are selected and classifier performance evaluation indicators are defined.After the two main study methods analyzed and evaluated,based on decision tree and high confidence level integrated prediction algorithm of support vector machine(SVM)is proposed.The support vector machine(SVM)technology is used to improve the prediction accuracy,and high degree of confidence the decision tree technology is used to reduce the prediction error of the mean and variance.Simulation results show that the proposed algorithm can both in prediction accuracy and confidence have good balance,to predict time is short,strong robustness,high prediction accuracy.This algorithm can provide reference information for the user to select suppliers.(3)On the basis of above theoretical research results,considering the practical application of supplier evaluation management requirements,the system function framework and system technology framework are designed.Combined with actual application demand preliminary supplier assessment management prototype system is achieved,verified the feasibility of proposed method.