R Language Feature Selection in R -- Removing Extraneous Features Removing features with zero or near-zero variance
Example
A feature that has near zero variance is a good candidate for removal.
You can manually detect numerical variance below your own threshold:
data("GermanCredit")
variances<-apply(GermanCredit, 2, var)
variances[which(variances<=0.0025)]
Or, you can use the caret package to find near zero variance. An advantage here is that is defines near zero variance not in the numerical calculation of variance, but rather as a function of rarity:
"nearZeroVar diagnoses predictors that have one unique value (i.e. are zero variance predictors) or predictors that are have both of the following characteristics: they have very few unique values relative to the number of samples and the ratio of the frequency of the most common value to the frequency of the second most common value is large..."
library(caret)
names(GermanCredit)[nearZeroVar(GermanCredit)]
