Intelligent decision support system for risk management in supply chains

Risk management is paramount for managing risks in supply chains. Although supply chain risk management involves identifying, assessing and predicting uncertain events, no particular strategy has been developed that considers both internal and external risk activities related to an organisation while managing it. This PhD thesis focuses on developing an intelligent decision support system that can identify essential risk labels related to operations, the features of service level agreement related to them and estimate risk levels based on analysing both the internal and external risk events in supply chains. This thesis provides the following three significant contributions to research on developing an intelligent decision support system for risk management in different supply chains. The first major contribution is the collection and analysis of the supply chain data obtained from different firms, with historical data and expert views used to determine the risk labels and evaluate the significance of features and their relation to the risk labels. More importantly, information from service level agreements and experts are used to create data sources for training and validating association rule mining and fuzzy inference techniques. The second significant contribution is that the critical features of risk labels are selected from data sources using emerging association patterns, which is integrated with association rule mining to determine the information related to the risk labels. The results show that this mechanism is better than other approaches for predicting risk labels from the data sources of supply chains. The third key contribution is the development of a fuzzy inference decision support system that can quantify the occurrence of risk labels based on the features selected using the emerging association patterns mechanism. The results obtained from its application on the collected data show that it outperforms existing techniques as it can accurately measure the severity of risk levels related to the risk labels and assist organisations in mitigating risk events and improving their market profits