Forecasting#
The sktime.forecasting
module contains algorithms and composition tools for forecasting.
Base#
|
Forecasting horizon. |
Naive#
|
Forecast based on naive assumptions about past trends continuing. |
|
Compute the prediction variance based on a naive strategy. |
Trend#
|
Trend based forecasts of time series data. |
|
Forecast time series data with a polynomial trend. |
|
Implements STLForecaster based on statsmodels.tsa.seasonal.STL implementation. |
Exponential Smoothing#
|
Holt-Winters exponential smoothing forecaster. |
|
ETS models with both manual and automatic fitting capabilities. |
ARIMA#
|
Automatically discover the optimal order for an ARIMA model. |
|
An ARIMA estimator. |
StatsForecast#
|
StatsForecast AutoARIMA estimator. |
Theta#
|
Theta method for forecasting. |
BATS/TBATS#
|
BATS forecaster for time series with multiple seasonality. |
|
TBATS forecaster for time series with multiple seasonality. |
Croston#
|
Croston’s method for forecasting intermittent time series. |
Prophet#
|
Prophet forecaster by wrapping Facebook’s prophet algorithm [R995275cbd543-1]. |
Unobserved Components#
|
Wrapper class of the UnobservedComponents model from statsmodels. |
Composition#
|
Forecast each series with separate forecaster. |
|
Ensemble of forecasters. |
|
Automatically find best weights for the ensembled forecasters. |
|
StackingForecaster. |
|
Meta-estimator for forecasting transformed time series. |
|
Pipeline for forecasting with exogenous data. |
|
Direct reduction from forecasting to tabular regression. |
|
Direct reduction from forecasting to time-series regression. |
|
Multioutput reduction from forecasting to tabular regression. |
Multioutput reduction from forecasting to time series regression. |
|
|
Recursive reduction from forecasting to tabular regression. |
Recursive reduction from forecasting to time series regression. |
|
|
Dir-rec reduction from forecasting to tabular regression. |
|
Dir-rec reduction from forecasting to time-series regression. |
|
MultiplexForecaster for selecting among different models. |
|
Make forecaster based on reduction to tabular or time-series regression. |
Online Forecasting#
|
Online Updating Ensemble of forecasters. |
|
Parameter free hedging algorithm. |
|
Ensemble forecasts with Non-negative least squares based weighting. |
Model Selection#
|
Cutoff window splitter. |
|
Single window splitter. |
|
Sliding window splitter. |
|
Expanding window splitter. |
|
Perform grid-search cross-validation to find optimal model parameters. |
|
Perform randomized-search cross-validation to find optimal model parameters. |
|
Split arrays or matrices into sequential train and test subsets. |
Model Evaluation (Backtesting)#
|
Evaluate forecaster using timeseries cross-validation. |
VAR (Vector Autoregression)#
|
A VAR model is a generalisation of the univariate autoregressive. |
SARIMAX#
|
SARIMAX forecaster. |