Creating dummy variables

tombishop · 18 May 2021 15:00

Hi, I am a bit rusty on some of the new DataSHIELD functionality. What is the best method to convert a factor variable (i.e. categorical with multiple levels) into dummy binary variables?

I can see there is probably a long-hand method with ds.boole, but wondered if there is a quicker way?

Thanks

demetris.avraam · 18 May 2021 15:40

Hi Tom,

You can create the dummy variables from a categorical variable using the ds.asFactor function and setting the argument fixed.dummy.vars equal to TRUE

tombishop · 18 May 2021 16:05

Hi Demetris, that is absolutely wonderful! Thank you!

tombishop · 19 May 2021 08:14

If I do this to multiple variables, then they all have column names like DV2 DV3 etc so there is conflict. What is the easiest way to rename columns in a matrix/dataframe, please?

demetris.avraam · 19 May 2021 08:47

Tom, the only way to rename variables using the current version of dsBaseClient is to use the ds.make function to replicate the variables with a different name and then add those variables back to your dataframe and remove the variables with the DV2 and DV3 names.

An easiest way is to use Tim’s “dh.renameVars” function that is included in the dsHelper package (this is a client-side function!)

github.com

lifecycle-project/ds-helper/blob/master/R/rename-vars.R

#' Rename multiple variables at once
#'
#' This function allows you to rename multiple variable from a dataframe. At the
#' moment it doesn't "rename" as such, it creates duplicate variables with the
#' new names. I've left it like this to keep in the spirit of ds/opal set up
#' by not automating the deletion of variables.
#'
#' @param conns connections object for DataSHIELD backends
#' @param df dataframe
#' @param names a dataframe or tibble containing two columns: "oldvar" (existing
#'              variable name), "newvar" (new variable name). Each row
#'              corresponds to one variable you want to rename
#'
#' @return None. The new variables are added to the df specified
#'
#' @importFrom dsBaseClient ds.assign ds.dataFrame
#' @importFrom purrr map pmap
#' @importFrom DSI datashield.connections_find
#'
#' @export

This file has been truncated. show original

tombishop · 19 May 2021 08:56

Thanks again!

Actually I just found another way, which is to use ds.matrixDimnames on the output from ds.asFactor (which is conveniently a matrix)

Topic		Replies	Views
Working with Logicals Beginner Support	1	194	2 December 2022
ds.dataFrame - Accessing variables Analyst Support	5	525	25 May 2020
How to compare variables of different tables Beginner Support	4	256	20 November 2022
Conversion of factors to numerics New functionality	0	556	12 November 2019
Changing reference level of variable not reflected in ds.glm result Analyst Support	2	140	27 September 2023

Creating dummy variables

Related topics