multivariate_selection: Variable selection for a multivariate logistic regression...
In TanguyPerennec/Autostats: Automated medical science statistics

Variable selection for a multivariate logistic regression model

multivariate_selection(
  DF,
  y,
  explicatives,
  keep,
  principal_factor = FALSE,
  method = "backward",
  criteria = "deviance",
  check_interactions = FALSE,
  alpha = 0.05,
  verbose = TRUE,
  delta = 7
)

`DF`	dataframe
`y`	character : variable to explain by 'explicatives'
`explicatives`	vector of character : variables to be selected in the multivariate model
`keep`	character or vector (optional) : variables that will be kept in the model no matter of its statistical importance Every variable known in the litterature to have interaction with y or other 'keep' variable should be listed in 'keep'.
`principal_factor`	(optional) : principal criteria to explain y. It will be th first variable to be selected and won't be removed from selection.
`method`	character : can be backward, forward. You can pass other arguments such as augmented or stepwise as follow "backward stepwise augmented".
`criteria`	character : the criteria that will be used to select models. Can be deviance, AIC and BIC.
`check_interactions`	logical : wether interaction between the principal_factor and the other variables should be checked.
`alpha`	num : the threesold that should be used to consider a test as significant
`verbose`	logical : whether the details of the calculs should be displayed on the console
`delta`	num : threeshold of difference in BIC or AIC to consider a model as more informative