What Md demonstrated

Built an end-to-end turbine yield predictor using leakage-free sensor inputs and a genuine time-based holdout. Compared a LinearRegression baseline with a tuned RandomForest, evaluated on held-out MAE/RMSE/R2, and documented the model’s intended use, limits, and drift-monitoring plan in a concise model card.

• Leakage-Free Feature Selection
• Honest Held-Out Evaluation
• Model Comparison and Tuning
• Plain-Language Model Communication