What Ambalavanan demonstrated

Built a leakage-aware turbine yield predictor end to end: framed TEY as the target, excluded downstream CO/NOx emissions, and validated on a genuine time-based holdout. Compared a linear baseline with a RandomForest, then reported held-out MAE, RMSE, R2, and residual behavior to show the model’s accuracy and limits in operational terms.

• Leakage-Free Feature Selection
• Time-Based Holdout Evaluation
• Model Comparison Against Baseline
• Residual-Based Error Reading