cvForecastControl: Default Cross-validation control

In evandeilton/cvforecast: Multiple forecasts based on CrossValidation

Description

Default Cross-validation control

Usage

cvForecastControl(stepSize = 1, maxHorizon = 1, minObs = 7,
  fixedWindow = TRUE, summaryFunc = tsSummary, preProcess = FALSE,
  ppMethod = "guerrero", cvMethod = "MAPE", tsfrequency = "month",
  OutlierClean = TRUE, residlevel = 0.1,
  dateformat = "%d/%m/%Y %H:%M:%S")

Arguments

stepSize	size step for the cross-validation samples
maxHorizon	forecasting horizon
minObs	minumum number of observation. Default is two times cycle of data
fixedWindow	keep fixed the sampling window, default is TRUE
summaryFunc	extra function to compute statistics of the model
preProcess	if TRUE does Box-Cox data transformation in to the data
ppMethod	if 'preProcess' is TRUE make 'guerrero' or 'loglik' tranformation. See BoxCox.lambda
cvMethod	accuracy method for best model choice. See accuracy
tsfrequency	time series data frequency
OutlierClean	if TRUE, remove outliers from the data. See tsclean
residlevel	confidence level for residual tests
dateformat	date format for charater dates

Value

list of parameters

Author(s)

LOPES, J. E.

Examples

# Control
myControl <- cvForecastControl(
minObs = 14,
stepSize = 5,
maxHorizon = 30,
summaryFunc=tsSummary,
cvMethod="MAPE",
tsfrequency='day',
OutlierClean=FALSE)
myControl

evandeilton/cvforecast documentation built on May 16, 2019, 9:36 a.m.