rg_fit_model: Linear model fitting with optional best subset selection,...

In sitscholl/Rgadgets: Contains various utility functions developed during the PhD

Description

This function allows you to fit a model between dependend and independend variables and calculate various selection routines.

Usage

rg_fit_model(
  predictors,
  response,
  method = "lm",
  ctrl = caret::trainControl(),
  subset = "full",
  preProc = c("center", "scale"),
  metric = ifelse(is.factor(response), "Accuracy", "RMSE"),
  maximize = ifelse(metric == "RMSE", FALSE, TRUE),
  tuneLength = 3,
  tuneGrid = NULL,
  verbose = T,
  withinSE = FALSE,
  minVar = 2,
  ...
)

Arguments

predictors	a data.frame with the predictor variables as columns
response	a vector with the response variable
method	string, type of statistical model for more details see caret::train()
ctrl	trainControl object
subset	string, subset selection routine that should be carried out. One of full, best, rfe or ffs
preProc	string or vector, pre processing routines that should be applied to predictors
metric	string, performance metric that is used to select the best model, defaults to RMSE for numeric and Accuracy for factor response
maximize	string, should metric be maximized?
tuneLength	numeric, see caret::train() for more details
tuneGrid	see caret::train() for more details
withinSE	Logical Models are only selected if they are better than the currently best models Standard error
minVar	Numeric. Number of variables to combine for the first selection
...	further arguments passed on to caret::train()

sitscholl/Rgadgets documentation built on Feb. 19, 2021, 1:24 a.m.