Panel Data¶
Yohou handles multi-series panel datasets natively. These notebooks cover every aspect of working with panel data: shared models with ColumnForecaster, per-series models with LocalPanelForecaster, panel preprocessing and stationarity, prediction intervals across groups, and panel cross-validation.
-
How to Configure LocalPanelForecaster
Wrap any forecaster with LocalPanelForecaster for fully independent per-group clones, parallel fitting via n_jobs, and selective group operations.
-
How to Forecast Multiple Columns Independently
Use ColumnForecaster to apply a point forecaster independently to each column of a multivariate time series.
-
How to Run Panel Cross-Validation
Time series cross-validation on panel data with GridSearchCV, selective group observation, rewind operations, and groupwise performance comparison.
-
How to Forecast Panel Data with ColumnForecaster
Apply a shared forecasting model across multiple series in a panel dataset using ColumnForecaster with the __ column separator convention.
-
How to Forecast Panel Prediction Intervals
Combine conformal and quantile regression intervals on panel data with per-group coverage analysis, calibration plots, and groupwise interval scoring.
-
How to Build Panel Feature Pipelines
Combine ColumnForecaster, FeaturePipeline, FeatureUnion, and DecompositionPipeline on panel data with per-group scoring on KDD Cup air quality.
-
How to Preprocess Panel Data
Automatic panel-aware transformation (StandardScaler, rolling stats, imputation) plus manual per-group workflows with get_group_df and dict_to_panel.
-
How to Use LocalPanelForecaster for Per-Series Models
Fit a separate reduction model for each panel series using LocalPanelForecaster and compare with a shared ColumnForecaster.
-
How to Apply Stationarity to Panel Data
Apply per-group stationarity transforms on panel data with SeasonalDifferencing, DecompositionPipeline (polynomial trend + pattern seasonality), and residuals.