Development of Proactive Risk-Predictive Model for 4PL Transaction Center Using PLS Regression and Neural Networks

Fourth-Party Logistics (4PL) transaction center aggregates trading partner competencies to provide comprehensive supply chain solutions. Since the 4PL transaction center deals with multiple category of trading partners, it offers both opportunities and risks. Especially, estimating risk in 4PL network involves collecting information from different combination of subjective and objective parameters which lacks predictive analytics. Hence, little work is carried out in synchronizing different metric scores to predict risk for managing transaction center effectively. In the first phase, risk assessment was carried out using Cormack's model. By combining individual scaling factors and probability arrived through request for information, risk probability index was estimated. Consequently, supply chain risk was determined considering total financial impact. Subsequently, risk evaluation of all trading partners with respect to high, moderate and low categories was performed utilizing prioritization matrix. In the second phase, predictive model was synthesized using Neural Network (NN) methodology. Moreover, optimal number of predictors was attained through Partial Least Square (PLS) regression. Finally, the NN was evaluated using verification dataset to ensure model adequacy. After achieving significant predictive accuracy, the developed model can be used by the coordinator to estimate risk proactively before conducting cross-segment integration. In addition, the model helps 4PL service provider to reduce supply disruption risks in the distribution network