API Reference
Data preparation
Prepare and validate historic, holiday, and temperature data.
- preprocessing.prepare_historic()
-
Prepare historic data.
- preprocessing.validate_historic()
-
Validate historic data.
- preprocessing.prepare_holidays()
-
Prepare holiday data.
- preprocessing.validate_holidays()
-
Validate holiday data.
- preprocessing.expand_holiday_windows()
-
Create a row for each holiday date - no lower/upper windows.
- preprocessing.prepare_temp()
-
Prepare midas air temperature data.
- preprocessing.interpolate_temp()
-
Interpolate missing air temperature data.
- preprocessing.validate_temp()
-
Validate midas air temperature data.
Data splitting
Create train/test splits and rolling forecast origin samples.
- splits.train_test_split()
-
Create a single train/test split.
- splits.rolling_forecast_origin()
-
Create rolling forecast origin train/test samples.
ARIMA
Configure and run ARIMA forecasts.
- arima.ARIMAParams
-
Parameters for the ARIMA model.
- arima.arima()
-
Fit ARIMA model and generate forecast.
Prophet
Configure and run Prophet forecasts.
- prophet.ProphetRegressor
-
Configuration and data for a single Prophet regressor.
- prophet.ProphetParams
-
Parameters for the Prophet model.
- prophet.prophet()
-
Fit Prophet model and generate forecast.
Naive benchmark
Configure and run Seasonal Naive forecasts.
- naive.SNaiveParams
-
Parameters for the SNaive model.
- naive.snaive()
-
Fit Seasonal Naive model and generate forecast.
Running forecasts
Run individual, multi-area/metric, and cross-validation forecasts.
- forecast.run_single_forecast()
-
Run a forecast for one metric and area.
- forecast.run_forecasts()
-
Run forecasts for every metric and area, sequentially or in parallel.
- forecast.run_cross_validation()
-
Run rolling forecast origin cross-validation.
Ensembles
Combine forecasts.
- ensemble.ensemble()
-
Calculate the mean of multiple forecasts.
Forecast evaluation
Calculate forecast accuracy measures.
- errors.forecast_errors()
-
Calculate forecast accuracy measures for a single metric/area/fold.
- errors.calculate_errors()
-
Calculate forecast accuracy measures for every metric/area/fold.
Plotting
Visualise results.
- plot.plot_forecast()
-
Plot historic data and forecast with 95% prediction intervals.
- plot.plot_cross_validation()
-
Plot observed and forecast values from all cross-validation folds.
- plot.plot_observed_against_forecast()
-
Scatter plot of observed values against forecast values.
- plot.plot_holiday_coverage()
-
Visualise when each holiday occurs over the whole time series.
- plot.plot_error_over_time()
-
Plot error over time.
- plot.plot_error_boxplot()
-
Plot cross-validation error distributions by forecast horizon.