Skip to content

Metrics

polars_ts.metrics

Metrics dataclass

mae(actual_col='y', predicted_col='y_hat', id_col=None)

Mean Absolute Error. See :func:polars_ts.metrics.forecast.mae.

rmse(actual_col='y', predicted_col='y_hat', id_col=None)

Root Mean Squared Error. See :func:polars_ts.metrics.forecast.rmse.

mape(actual_col='y', predicted_col='y_hat', id_col=None)

Mean Absolute Percentage Error. See :func:polars_ts.metrics.forecast.mape.

smape(actual_col='y', predicted_col='y_hat', id_col=None)

Symmetric MAPE. See :func:polars_ts.metrics.forecast.smape.

mase(actual_col='y', predicted_col='y_hat', id_col='unique_id', time_col='ds', season_length=1)

Mean Absolute Scaled Error. See :func:polars_ts.metrics.forecast.mase.

crps(actual_col='y', quantile_cols=None, quantiles=None, id_col=None)

CRPS (quantile approximation). See :func:polars_ts.metrics.forecast.crps.

lag_features(lags, target_col='y', id_col='unique_id', time_col='ds')

Create lag features. See :func:polars_ts.features.lags.lag_features.

rolling_features(windows, aggs=None, target_col='y', id_col='unique_id', time_col='ds', center=False, min_samples=None)

Create rolling features. See :func:polars_ts.features.rolling.rolling_features.

calendar_features(features=None, time_col='ds')

Extract calendar features. See :func:polars_ts.features.calendar.calendar_features.

fourier_features(period, n_harmonics=1, time_col='ds', id_col='unique_id')

Generate Fourier features. See :func:polars_ts.features.fourier.fourier_features.

log_transform(target_col='y')

Apply log1p transform. See :func:polars_ts.transforms.log.log_transform.

inverse_log_transform(target_col='y')

Invert log transform. See :func:polars_ts.transforms.log.inverse_log_transform.

boxcox_transform(lam, target_col='y')

Apply Box-Cox transform. See :func:polars_ts.transforms.boxcox.boxcox_transform.

inverse_boxcox_transform(lam=None, target_col='y')

Invert Box-Cox transform. See :func:polars_ts.transforms.boxcox.inverse_boxcox_transform.

difference(order=1, period=1, target_col='y', id_col='unique_id', time_col='ds')

Apply differencing. See :func:polars_ts.transforms.differencing.difference.

undifference(order=1, period=1, target_col='y', id_col='unique_id', time_col='ds')

Invert differencing. See :func:polars_ts.transforms.differencing.undifference.

expanding_window_cv(n_splits=5, horizon=1, step=1, gap=0, id_col='unique_id', time_col='ds')

Expand-window CV. See :func:polars_ts.validation.splits.expanding_window_cv.

sliding_window_cv(n_splits=5, train_size=10, horizon=1, step=1, gap=0, id_col='unique_id', time_col='ds')

Slide-window CV. See :func:polars_ts.validation.splits.sliding_window_cv.

rolling_origin_cv(n_splits=5, initial_train_size=None, horizon=1, step=1, gap=0, fixed_train_size=None, id_col='unique_id', time_col='ds')

Roll-origin CV. See :func:polars_ts.validation.splits.rolling_origin_cv.

conformal_interval(cal_residuals, coverage=0.9, residual_col='residual', predicted_col='y_hat', id_col=None, symmetric=True)

Add conformal prediction intervals. See :func:polars_ts.probabilistic.conformal.conformal_interval.