tsaggregate: Series Aggregation

In tsmodels/tsvets: Vector ETS Models

Series Aggregation

Description

Aggregates estimated, predicted or simulated series based on a weighting vector.

Usage

## S3 method for class 'tsvets.estimate'
tsaggregate(object, weights = NULL, return_model = FALSE, ...)

## S3 method for class 'tsvets.predict'
tsaggregate(object, weights = NULL, ...)

## S3 method for class 'tsvets.simulate'
tsaggregate(object, weights = NULL, ...)

Arguments

object	object of class “tsvets.estimate”. “tsvets.predict” or “tsvets.simulate”.
weights	vector of weights of length equal to the number of series.
return_model	if the estimated object is a homogeneous coefficients model with common lamda parameter (of either 0, 1 or NULL), then it will return a univariate tsvets object of class “tsvets.estimate”.
...	not currently used.

Details

For an estimated object which has common components (homogeneous coefficients) for all estimated states and common lambda parameter, then a reconstructed object of the same class representing the weighted representation of the model is returned. In all other cases and input classes, the returned object will depend on whether the lambda parameter was 0 or 1 for all underlying series. For the case of the log transform (lambda = 0), then the states aggregate (given a weighting vector) whilst the actual, fitted, predicted or simulated values aggregate in logs and are then exponentiated (a multiplicative model). In all other cases only the actual, fitted, predicted or simulated values are returned representing the weighted aggregation of the underlying series given the weighting vector.

Value

Depends on the input class.

References

Hyndman, Rob and Koehler, Anne B and Ord, J Keith and Snyder, Ralph D, 2008, Forecasting with exponential smoothing: the state space approach, Section 17.1.2,,Springer Science \& Business Media.

tsmodels/tsvets documentation built on June 13, 2022, 2:14 p.m.