What Hassan demonstrated

Built an end-to-end gas-turbine TEY predictor using leakage-free sensor inputs, a time-based train/test split, and a baseline LinearRegression versus a tuned RandomForest. Reported honest held-out performance (MAE 1.502 MWh, RMSE 1.908 MWh, R² 0.9835), documented why CO/NOx were excluded, and described the model’s limits, drift risks, and monitoring plan in a manager-friendly model card.

• Leakage-Free Feature Selection
• Time-Based Holdout Evaluation
• Honest Model Assessment
• Plain-Language Model Communication