Time Series Cross-Valiation

This is a package to automate the cross-validation of time-series models, particularly those created by the forecast and caret packages.

Install as follows:

devtools::install_github('zachmayer/cv.ts')

You can then cross-validate different time series models as follows:

data("AirPassengers")
x <- AirPassengers
myControl <- tseriesControl(maxHorizon=4)
theta_model <- cv.ts(x, thetaForecast, tsControl=myControl)
arima_model <- cv.ts(x, auto.arimaForecast, tsControl=myControl)
ets_model <- cv.ts(x, etsForecast, tsControl=myControl)

theta_model$results
arima_model$results
ets_model$results

This package has been off my radar screen for a long time, and I need to write lots of documentation and tests. Please add issues to the issue tracker, and I welcome any pull requests. https://github.com/zachmayer/cv.ts/issues