Skip to contents

Prepare ML helper function

Source: R/prepare_ml_vars.R
prepare_ml_vars.Rd

Identify variable sets from input matrix that might require extra steps in data preparation, e.g. skewed variables to be log transformed, counts

Usage

prepare_ml_vars(
  data,
  thres_count = NULL,
  thres_log = NULL,
  thres_lump = NULL,
  remove = c(".id", ".out", ".status", ".time")
)

Arguments

data

the data set to be searched for feature sets with specific characteristics relevant for further data preparation

thres_count

used to detect integer columns with up to thres_count distinct values (might be excluded from further processing, e.g. log & normalization)

thres_log

threshold for log transformation

thres_lump

proportion threshold for factor lumping; used to detect factors with exactly one level having a relative frequency below thres_lump

remove

columns to be excluded from all identified sets; defaults to c(".id", ".out", ".status", ".time")

Value

A list with slots specifying the detected variable sets of interest. NA if required thresholds were not defined; NULL if no variables meet the corresponding criteria.

count

assumed to be counts

log

to be log transformed as the skewness exceeds thres_log

nolump

to be excluded from lumping

Authors

Maike Ahrens (ahrensmaike), Sebastian Voss (svoss09)