Demand forecasting and rebalancing analysis of urban bikeshare systems across multiple cities, using riders' patterns to inform fleet distribution strategy.
-
Updated
Mar 27, 2026 - Jupyter Notebook
#
forecas
Here are 3 public repositories matching this topic...
Multi-model time series forecasting platform: SARIMA, LightGBM (SMAPE 12%), and N-BEATS (PyTorch, from scratch). Walk-forward validation across 8 expanding windows. Lag/rolling/Fourier feature engineering. 16 tests.
python machine-learning time-series pytorch lightgbm statsmodels sarima forecas nbeats walk-forward-validation retail-forecasting
-
Updated
Jul 22, 2026 - Python
Add this topic to your repo
To associate your repository with the forecas topic, visit your repo's landing page and select "manage topics."