Machine Learning Algorithm For Supplier Classification

Supplier classification plays a critical role in the supply chain management,and it is the premise of formulating supplier management strategies.At present,the classification standards for suppliers are not the same,so this paper reviews a variety of classification standards and finds that the Kraljic matrix is a widely recognized classification standard.In addition,there are many research methods for supplier classification.One of the most commonly used evaluation method is the multi-criteria decision-making method,but this method cannot avoid human subjective judgment.If there are too many evaluation criteria,it will increase the complexity of the evaluation process.For a comprehensive and objective assessment process,we propose the use of supervised machine learning algorithms to classify various suppliers.First,the author implements the Kraljic matrix model to identify the key index factors that affect supplier classification from the two dimensions of profit and risk.Second,supervised machine learning algorithms are used to divide suppliers into four categories,strategic suppliers,leverage suppliers,bottleneck suppliers,general suppliers,and performance of models are validated.This paper uses a total of five supervised learning algorithm models,namely support vector machine,K-nearest neighbor,Naive Bayes,decision tree and random forest,and compares the performance of the five classification models to find the best,which is the innovation of this paper.At the same time,PCA(principal component analysis)method is also applied to confirm the key index factors that affect supplier classification,which can be references for future supplier selection.Finally,this paper proposes corresponding management strategies according to different types of suppliers.