Synergy between AI, HCI, and Operations Research: An Interactive Decision Support System for Risk Mitigation under Uncertainty

Decision-making in Operations Research (OR) unfolds amid high uncertainty, demanding forecasting tools that are both accurate and interpretable. This study proposes a five-layer framework that couples Deep Learning with Human-Computer Interaction (HCI) principles to serve as a Decision Support System (DSS) for commodity price risk. Four GluonTS architectures (DeepAR, TFT, Simple Feed Forward, and DeepNPTS), tuned with Optuna, were evaluated on the Brent, WTI, and OPEC daily benchmarks from 2016 to 2024 through a composite score over RMSE, SMAPE, and MASE. DeepNPTS achieved the highest composite score across all three series, with the lowest SMAPE in each case. The contribution lies in the interaction layer: coordinated visual views and a human-in-the-loop path that enable decision-makers to compare models and rerun scenarios, converting opaque forecasts into auditable inputs for CAPEX, hedging, and procurement decisions.