A framework which provides instruments for the time series forecasting tasks and strategies.
- Local-modelling:
- Individual model for each time series.
- Global-modelling:
- One model for all time series;
- features for individual observations do not overlap in the context of different time series.
- Multivariate-modelling:
- One model for all time series;
- features for observations corresponding to the same time point are concatenated for all time series, and the output of the model for a single observation from the test sample is a vector of predicted values whose length is equal to the number of time series under consideration.
- Recursive:
- one model for all points of the forecast horizon;
- training: the model is trained to predict one point ahead;
- prediction: a prediction is iteratively made one point ahead, and then this prediction is used to further shape the features in the test data.
- Recursive-reduced:
- one model for all points in the prediction horizon;
- training: the model is trained to predict one point ahead;
- prediction: features are generated for all test observations at once, unavailable values are replaced by NaN.
- Direct:
- An individual model for each point in the prediction horizon.
- DirRec:
- An individual model for each point in the prediction horizon.
- learning and prediction: iteratively builds test data for an individual model at each step, makes a prediction one point ahead, and then uses this prediction to further generate features for subsequent models.
- MultiOutput (MIMO - Multi-input-multi-output):
- one model that learns to predict the entire prediction horizon (the model output for one observation from the test sample is a vector of predicted values whose length is equal to the length of the prediction horizon).
- FlatWideMIMO:.
- mixture of Direct and MIMO, fit one model, but uses deployed over horizon Direct's features.