checknzv: Check nzv

In sunsiyu/featurer: Bag of Tricks to Ease Feature Engineering

Description

This function is needed also because I would like to have the insight of what nzv can contribute to feature selection / filtering. Some interesting topics: do we always want to drop zv? when do we want to drop some nzv also?

Usage

checknzv(df, the_nzv = NULL, n = min(10, round(ncol(df)/10)),
  v_label = f_train_label, verbose = T, ...)

Arguments

df	the dataset as a data.frame or matrix
the_nzv	default NULL, the already calculated nzv by caret::nzv
n	first n features with high nzv
v_label	vector of lable
verbose	logical
...	other parameters

Value

a list of features with high nzv checks

sunsiyu/featurer documentation built on May 30, 2019, 8:39 p.m.