subtopic

Score on the test period and compare

Predict the whole test period, undo the scaling, and compute MAE, RMSE and MAPE. Put LSTM, seasonal naive and LightGBM in one table. Plot forecast vs actual for 2 weeks and the average error by hour of day. Write a short decision: which model would you deploy and why?

Builds on: S02 · Build: An Evaluation Harness (used for every model)