About One hot encoder

[d-ghale]

Is there a reason why "one_hot_encoder" creates the columns in integer format instead of logical ?

I need help to find an efficient way to convert the columns I created using "one_hot_encoder" into lgl format without expliciting writing out the list of columns.

Thanks!

[ELToulemonde]

Hi,

Hmm... I don't really know. Why not? Do you have a use case where 0 and 1 are painful?

[d-ghale]

Yes, say I create a numeric column called total_count that contains 0 or 1. I use one_hot_encoder to create a logical column but it is expressed as 0 or 1, not TRUE or FALSE. When I use whichAreInDouble I get a false result saying these two columns are the same.

[ELToulemonde]

Ok I say

• First, I don't realy see why it is a problem, since both columns contains exactly the same information (for example function whichAreBijection will drop it even if it is TRUE/FALSE or 0/1.

• Second, here is a dirty fix; pass the column you wish to fix into argument keep_cols in whichAreDouble

• Last, I added the argument type into one_hot_encoder, this argument let's you choose between integer, numeric and logical for the type of the result. It will be released in future CRAN release.

Until CRAN release, you can have this feature by installing package directly from github

library(devtools)
install_github("ELToulemonde/dataPreparation")

Hope I answered your question.

If you have any other question/remarks please don't hesitate.

I close.

Emmanuel-Lin

[d-ghale]

Thanks! Yes, whichAreBijection will drop it even if it is TRUE/FALSE or 0/1. I am using fastFilterVariables with type = 2 to drop constant and double columns. Your update solved my problem.

Thank You!