Machine Learning and Decision Analysis Approach to Optimal Reorder Policy for Critical parts in Manufacturing

Chew Man Ting

Abstract

An efficient and effective inventory management system greatly reduces the control of inventory level and fulfill customer demand timely. This thesis focuses on the need for reducing the cost incurred due to a shortage of parts and increasing customer satisfaction by reducing machine downtime costs. Keeping sufficient inventory without sacrificed customer service satisfaction is one key business challenge to all the companies. A case study research approach is adopted for this study. Data collection through the interviewing several stakeholders, examine internal documentation and information of AM company and Customer A company. The outcome of this study is to propose a better reorder point model of critical parts for a AM company’s production equipment in the manufacturing plant. The research approach includes: classify production equipment components, cluster the critical part, analyze and define the reorder point models. The details of the study are illustrated by the analytic hierarchy process (AHP), machine learning method and decision analysis in terms of theories and mathematical formulas. Through the analysis, AM company is expected to have a cost-saving of $1 million per year in emergency shipping support while Customer A is expected to avoid a loss of $800,000 due to machine downtime by adapting the recommended reorder point model. This thesis also helps to facilitate company management on refining the inventory management from the stock reorder level viewpoint.