A practical introduction to time series analysis and forecasting using Python, illustrated with real climate data from the German Weather Service (DWD).
Contents¶
| Notebook | Topic |
|---|---|
| 00 - Load DWD Data | Fetching and preparing weather station data |
| 01 - Time Series Properties | Stationarity, seasonality, autocorrelation |
| 02 - Traditional Methods | ARIMA, exponential smoothing, classical decomposition |
| 03 - ML & Deep Learning | Tree-based and deep learning methods with Darts |
| 04 - AutoML | Automated model selection with AutoGluon |
| 05 - Transformers | Temporal Fusion Transformer with Darts |
Data¶
All examples use hourly/daily temperature and precipitation records from five DWD climate stations across Germany.
How to run¶
Install uv
uv sync
uv run jupyter lab