Modeling and Optimization of a Supply Chain Loop's Performance by an Integrated Neural Network-Fuzzy Regression-Ridge Regression Approach

The goal of this research is to identify the significant factors affecting the firm performance and estimate the system behavior in different operating conditions. By determining the statistical relations of the productivity and effectiveness of the firm with these factors, a decision-making framework can be provided to improve the system performance within the competitive strategy of the whole supply chain. This research presents a flexible meta modeling approach for modeling and optimization the operating performance of a firm in a supply chain by integrating Fuzzy Linear Regression (FLR), Ridge Regression (RR), and Artificial Neural Network (ANN). The efficiencies of FLR, RR and ANN approaches in prediction and modeling are compared and the superior approach is selected according to Mean Absolute Percentage Error (MAPE) and minimum number of observation (n) for test data calculated from OC curve.