Integrated Model for Predicting Supply Chain Risk Through Machine Learning Algorithms

Abstract

The machine learning model has become a critical consideration in the supply chain. Most of the companies have experienced vari-ous supply chain risks over the past three years. Earlier risk prediction has been performed by supply chain risk management. In this study, an integrated supply chain operations reference (ISCOR) model has been used to evaluate the organization's supply chain risk. Machine learning (ML) has become a hot topic in research and industry in the last few years. With this motivation, we have moved in the direction of a machine learning-based pathway to predict the supply chain risk. The great attraction of this research is that suppliers will understand the associated risk in the activity. This research includes data pre-processing, feature extraction, data transformation, and missing value replacement. The proposed integrated model involves the support vector machine (SVM), k near-est neighbor (k-NN), random forest (RF), decision tree (DT), multiple linear regression (MLR) algorithms, measured performance, and prediction of supply chain risk. Also, these algorithms have performed a comparative analysis under different aspects. Among the other algorithms, the random forest algorithm achieves an accuracy of 99% and has accomplished superior results with a maxi-mum precision of 0.99, recall of 0.99, and F-score of 0.99 with 1% error rate. The model’s prediction indicates that it can be used to find the supply chain risk. Finally, the limitation and the challenges discussed also provide an outlook for future research direction to perform effective management to mitigate the risk.