Logistics network forecasting & optimization

Research question

How should a logistics network respond when a depot closes? Using competition data for 81 depots and 1,049 directed routes, our team connected demand forecasting with capacity-constrained redistribution and network redesign.

My contribution

• Participated in the MathorCup team project linking time-series prediction, constrained optimization, and network evaluation.
• Worked within a modeling study organized around two depot-closure scenarios and the relationship between forecast demand, route capacity, and load balance.

Technical approach

• Organize historical freight volumes by route and date, address missing observations, and fit ARIMA models to forecast the following month’s demand.
• Formulate redistribution with flow consistency, capacity, and route-change considerations; use simulated annealing to search allocation plans.
• Extend the model to daily route opening and closure decisions, then use entropy-weighted TOPSIS to rank depots and routes by freight and connectivity indicators.

Results & outcomes

• Awarded First Prize in the 2023 MathorCup undergraduate division.
• The team report provides demand forecasts, redistribution plans for DC5 and DC9 closure scenarios, and a ranked assessment of network importance. These outputs connect predictive modeling with concrete network adjustment decisions.