DFML: DFML

In jdestefani/ExtendedDFML: Extended Dynamic Factor Machine Learner

Description

Core function implementing the EDFML technique. Takes the input data X, and passes it through the dimensionality reduction step (via the dimensionalityReduction fucntion) to obtaing the dynamic factors Z. Factors are then forecast (via the forecaster function), in order to obtain the factor forecasts (Z_hat). Finally, factor forecast are transformed to forecast in the original space via an inverse dimensionality reduction (dimensionality increase).

Usage

DFML(
  X,
  dimensionality_method = DIMENSIONALITY_METHODS,
  forecast_family = FORECAST_FAMILY,
  dimensionality_parameters,
  forecast_parameters,
  components,
  h
)

Arguments

X	nxN matrix containing the N time series as columns, each one of length n time steps
dimensionality_method	Dimensionality reduction method to employ - String among those defined in DIMENSIONALITY_METHOD
forecast_family	Forecasting family of method to employ - String among those defined in FORECAST_FAMILY
dimensionality_parameters	Parameters to be passed to the dimensionalityReduction function - List
forecast_parameters	Parameters to be passed to the forecaster function - List
components	Number of desired factors - numeric scalar
h	Forecasting horizon - numeric scalar

Value

List containing:

• X_hat: hxN matrix containing the forecasts of N time series as columns, each one of length h time steps

• Model_dim: Dimensionality reduction model as produced by dimensionality_reduction function

• Time_dim: Computational time required to run the dimensionality reduction model - numeric scalar

• Time_forecast: Computational time required to run the forecasting model - numeric scalar

Examples

#See tests/testthat directory on https://github.com/jdestefani/ExtendedDFML

jdestefani/ExtendedDFML documentation built on Dec. 20, 2021, 10:04 p.m.