In this article, we will explain how you can convert multiple columns (variables) to factor in R using both base R and dplyr packages.
In R, categorical variables need to be set as factor variables. Some of the numeric variables which are categorical in nature need to be transformed to factor so that R treats them as a grouping variable.
Let's create a sample data frame called mydata
having 5 variables.
# Create a dummy data frame mydata <- data.frame( var1 = c("A", "B", "C"), var2 = c("X", "Y", "Z"), var3 = c(1, 2, 3), var4 = c(7, 8, 9), var5 = c("G", "H", "I") )
How to Convert Numeric Columns to Factor
In the dataframe named 'mydata', we have two numeric columns 'var3' and 'var4'. We do not want to explicitly name these columns while converting them to factor.
is.numeric
function is used to identify the numeric columns. Then, the as.factor
function is applied to convert the columns to factors.
Base R
numeric_cols <- sapply(mydata, is.numeric) mydata[numeric_cols] <- lapply(mydata[numeric_cols], as.factor) str(mydata)
In base R, you can convert multiple columns (variables) to factor using lapply
and sapply
functions. The lapply and sapply functions are used to perform multiple iterations (loops) in R. The only difference between them is that lapply returns list. Whereas sapply returns vector or matrix.
dplyr
library(dplyr) mydata <- mydata %>% mutate(across(where(is.numeric), as.factor)) str(mydata)
In dplyr package, the across
function allows you to apply a transformation across multiple columns. The mutate
function is used to modify the columns of a dataframe.
How to Convert All Columns to Factor
names(mydata)
command returns a character vector containing the names of all the columns in the dataframe named "mydata".
Base R
col_names <- names(mydata) mydata[,col_names] <- lapply(mydata[,col_names] , factor) str(mydata)
dplyr
library(dplyr) col_names <- names(mydata) mydata <- mydata %>% mutate(across(all_of(col_names), as.factor)) str(mydata)
Converting Columns to Factor using Column Position
In this case, we are converting first, second, third and fifth variables to factor variables. mydata is a data frame.
Base R
names <- c(1:3,5) mydata[,names] <- lapply(mydata[,names] , factor) str(mydata)
dplyr
library(dplyr) names <- c(1:3, 5) mydata <- mydata %>% mutate(across(names, as.factor)) str(mydata)
Converting Columns to Factor using Column Names
In this case, we are converting two columns 'var2' and 'var5' to factor variables.
Base R
names <- c('var2' ,'var5') mydata[,names] <- lapply(mydata[,names] , factor) str(mydata)
dplyr
library(dplyr) names <- c('var2', 'var5') mydata <- mydata %>% mutate(across(names, as.factor)) str(mydata)
Convert Columns to Factor Based on Condition
Suppose you want to convert only those columns to factors which have a number of unique values less than 4.
Base R
col_names <- sapply(mydata, function(col) length(unique(col)) < 4) mydata[ , col_names] <- lapply(mydata[ , col_names] , factor)
dplyr
library(dplyr) col_names <- sapply(mydata, function(col) length(unique(col)) < 4) mydata <- mydata %>% mutate(across(names(col_names)[col_names], as.factor)) str(mydata)
wow.... thank you so much for this. i've been searching for this all over the internet and finally found it here...
ReplyDeleteeven i
Deletethese do not work
ReplyDeleteThank you, it was what I was looking for!
ReplyDeleteI believe that in 5. the right code for col_names is:
ReplyDeletecol_names <- sapply(mydata,
function(col) {length(unique(col) < 4} )
I followed your data type conversion example on my Excel ".xlsx" file. The numeric columns were converted into factors which is required by the package that I am using. However, when I run the R package, I get an error that goes like this: Error in '$<- .data.frame.'(*tmp*', "Trt", value = character(0)) replacement has 0 rows, data has 20.
ReplyDeleteWhen I check the data type conversion using str() function, the numeric columns were converted to factors as I desired. However, it seems that the "myData[, names]" statement did not capture any of the data rows in the dataframe when in fact it should.
Any helpful thoughts about my problem?
Thank you.
Hi
ReplyDeleteCan you please clarify that variables like exposuretime, size, concentration should be included in the generalized linear model as numeric or factors? Thanks