multiFC: Process multiple models and produce forecasts with summaries

In MichaelFolkes/forecastR_package: Forecast stuff

Description

Process multiple models and produce forecasts with summaries

Usage

multiFC(
  data.file,
  settings.list,
  do.retro = FALSE,
  retro.min.yrs = 15,
  out.type = c("short", "full"),
  int.type = c("None", "Retrospective", "Prediction", "Bootstrap"),
  int.n = 100,
  boot.type = "meboot",
  tracing = FALSE
)

Arguments

data.file	The data file read in with read.csv.
settings.list	A list with list sub-elements. See details.
do.retro	A Boolean (default is FALSE). Run a retrospective
retro.min.yrs	An integer of length one. Default is 15.
out.type	A character vector of length one. The choices are: "short" or "full". See value for details.
int.type	Either "None","Retrospective","Prediction", or "Bootstrap"
int.n	number of samples used for the interval calculations
boot.type	= "meboot" or "stlboot", only applies if int.type = "Bootstrap"
tracing	A Boolean (default is FALSE)

Details

The settings.list argument is a list defining each forecast model in separate lists that can be uniqely named. Each model list must have the elements "model.type" and "settings". The settings element is also a list. For example: settings.list= list(Naive1 = list(model.type="Naive",settings=list(avg.yrs=1)), SibRegSimple = list(model.type="SibRegSimple",settings=NULL))

Value

A list is produced. If argument out.type = "short" and do.retro = FALSE, then only generate a summary table of point forecasts. If out.type = "short" and do.retro= TRUE, then generate 3 versions of retrospective summary and array of fitted performance measures. If out.type ="full" and do.retro= TRUE, then also store the model fits for each retro year. If out.type ="short" and = TRUE, then TBD. If out.type ="full" and do.boot= TRUE, then TBD

MichaelFolkes/forecastR_package documentation built on April 4, 2021, 5:14 a.m.