A final exam notebook covering both the theory and the practice of time series forecasting — ARIMA/SARIMA, forecast error metrics, baseline models, and gradient-boosted forecasting with LightGBM.
This notebook is split into two parts. Part 1 is a set of applied essay questions: differentiating ARIMA from SARIMA, comparing forecast error metrics, distinguishing mean forecasts from naive forecasts, explaining LightGBM for time series forecasting, and justifying the train/test split in a forecasting context. Part 2 is a hands-on forecasting implementation, applying those concepts directly to a time series dataset.