📄️ Time series forecasting
Forecast daily pharmacy demand at 28 days: baselines, ARIMA, Prophet, gradient boosting and temporal Transformers, with rolling validation and honest intervals.
📄️ 1. Series vs table
Module 1 of the Time Series Forecasting premium course: temporal dependence, horizon, frequency, the split without shuffling, future leakage and vocabulary.
📄️ 2. Decomposition
Module 2 of the Time Series Forecasting premium course: additive versus multiplicative decomposition, STL, multiple seasonalities, and how to read the residuals.
📄️ 3. Stationarity and ACF
Module 3 of the Time Series Forecasting premium course: ADF test, simple and seasonal differencing, ACF and PACF, reading them to pick orders.
📄️ 4. Baselines
Module 4 of the Time Series Forecasting premium course: naïve, seasonal naïve, moving average and drift, and why sophisticated models must clearly beat them.
📄️ 5. ARIMA and SARIMA
Module 5 of the Time Series Forecasting premium course: AR, I, MA components, seasonal orders, AIC selection, residual diagnostics and forecasts with intervals.
📄️ 6. Smoothing and Prophet
Module 6 of the Time Series Forecasting premium course: Holt-Winters, Prophet components, holidays, changepoints and the honest limits of Prophet in practice.
📄️ 7. Features and regressors
Module 7 of the Time Series Forecasting premium course: lags, rolling windows, day-of-week, holidays, promotions and gradient boosting on a supervised table.
📄️ 8. Deep approaches
Module 8 of the Time Series Forecasting premium course: windowing, LSTM, temporal Transformers, foundation models, and when deep learning wins or loses.
📄️ 9. Rolling validation
Module 9 of the Time Series Forecasting premium course: rolling origin, multiple horizons, MAE, RMSE, MAPE at zero, MASE, and interval evaluation.
📄️ 10. Project: demand forecast
Module 10 of the Time Series Forecasting premium course: final results table, argued model choice, quantile intervals, retraining plan and communicating uncertainty.
📄️ Recap and exam
Complete recap of the Time Series Forecasting premium course: baselines, ARIMA, Prophet, gradient boosting, deep models, rolling validation, and the 40-question exam.