ForeCA (version 0.2.8-1)

ForeCA-package: Implementation of Forecastable Component Analysis (ForeCA)

Description

Forecastable Component Analysis (ForeCA) is a novel dimension reduction technique for multivariate time series \(\mathbf{X}_t\). ForeCA finds a linar combination \(y_t = \mathbf{X}_t \mathbf{v}\) that is easy to forecast. The measure of forecastability \(\Omega(y_t)\) (Omega) is based on the entropy of the spectral density \(f_y(\lambda)\) of \(y_t\): higher entropy means less forecastable, lower entropy is more forecastable.

The main function foreca runs ForeCA on a multivariate time series \(\mathbf{X}_t\).

Consult NEWS.md for a history of release notes.

Author

Maintainer: Georg M. Goerg [email protected]

References

Goerg, G. M. (2013). “Forecastable Component Analysis”. Journal of Machine Learning Research (JMLR) W&CP 28 (2): 64-72, 2013. Available at https://proceedings.mlr.press/v28/goerg13.html.

See Also

Useful links:

• https://github.com/gmgeorg/ForeCA

• Report bugs at https://github.com/gmgeorg/ForeCA/issues

Examples

XX <- ts(diff(log(EuStockMarkets)))
Omega(XX)

plot(log10(lynx))
Omega(log10(lynx))

if (FALSE) {
ff <- foreca(XX, n.comp = 4)
ff
plot(ff)
summary(ff)
}