CODEWITHSUNDEEP
rgroup-bydplyr
cgxytf
cgxytf

Reputation: 431

Calculate proportion (percent) for each column of a dataset

I'm trying to calculate the proportion (percent) of categories in each column of a dataset.

Example data:

df <- data.frame(
    "Size" = c("Y","N","N","Y","Y"), 
    "Type" =  c("N","N","N","Y","N"), 
    "Age" = c("N","Y","N","Y","N"), 
    "Sex"=c("N","N","N","N","N")
  )

df

Data produces a table like this:

    Size Type Age Sex
1    Y    N   N   N
2    N    N   Y   N
3    N    N   N   N
4    Y    Y   Y   N
5    Y    N   N   N

I've tried using prop.table to calculate proportion for one category: 

prop.table(table(df$Size))

This works, but only calculates the percent of Y or N answers for one column. This dataset is quite large, so I'd like to calculate the proportion for each category at once.

My goal is to have a table that shows the proportion of "yes" answers for each column.

Like this:

       Proportion Y
Size    0.60
Type    0.20
Age     0.40
Sex     0.00

I am relatively new to R, so any help would be appreciated!

Upvotes: 2

Views: 1040

Answers (2)

StupidWolf
StupidWolf

Reputation: 46978

A dplyr approach:

library(dplyr)
df %>% summarise_all(~mean(.=="Y"))

If you have more than one group:

df1 = data.frame(class="A",df)
df2 = data.frame(class="B",df)
#make df2 different
df2$Size<- rep("Y",5)
newdf = rbind(df1,df2)
newdf %>% group_by(class) %>% summarise_all(~mean(.=="Y"))

Upvotes: 2

Ronak Shah
Ronak Shah

Reputation: 389315

One way in base R would be to use apply column-wise on a logical vector 

apply(df == "Y", 2, mean)

#Size Type  Age  Sex 
# 0.6  0.2  0.4  0.0

A simpler version with colSums.

colMeans(df == "Y")

Upvotes: 3

Related Questions