Guides ยท Science
Time Series Forecasting Basics
Build simple time series forecasts
This guide covers prepping time series data, trying baseline models (naive, seasonal naive), moving to ARIMA/ETS or simple ML, and evaluating with rolling-origin cross-validation.
- time series
- forecasting
- arima
- ets
- cross-validation
Prepare the series
Handle missing values, outliers, and ensure consistent frequency.
Start with baselines
Use naive and seasonal naive to set a floor before complex models.
Try simple models
Test ARIMA/ETS or Prophet-like models; avoid overfitting.
Evaluate properly
Use rolling-origin CV and track MAE/MAPE; monitor drift after deployment.