R summarize unique values across columns based on values from one column

Question

I want to know the total number of unique values for each column based on the values of var_1.

For example:

Test <- data.frame(var_1 = c("a","a","a", "b", "b", "c", "c", "c", "c", "c"), var_2 = c("bl","bf","bl", "bl","bf","bl","bl","bf","bc", "bg" ), var_3 = c("cf","cf","eg", "cf","cf","eg","cf","dr","eg","fg"))

The results I am looking for would be based on the values in var_1 and should be:

var_1 var_2 var_3
a     2     2
b     2     1
c     3     4

However, after trying various methods (including apply and table) - aggregate has been the closest thing to what I am looking for, but this script results in a summary of the total number of entries for each value of var_1, but the total is not unique

agbyv1= aggregate(. ~ var_1, Test, length) 

var_1 var_2 var_3
a     3     3
b     2     2
c     5     5

I tried

unqbyv1= aggregate(. ~ var_1, Test, length(unique(x)))

but that didn't work.

Any help is greatly appreciated.

Answer 1

Try

library(dplyr)
Test %>%
      group_by(var_1) %>% 
      summarise_each(funs(n_distinct(.)))

Or

library(data.table)#v1.9.5+
setDT(Test)[, lapply(.SD, uniqueN), var_1]

If there are NAs

setDT(Test)[, lapply(.SD, function(x) uniqueN(na.omit(x))), var_1]

Or you can use aggregate. By default, the na.action=na.omit. So, we don't need any modifications.

aggregate(.~ var_1, Test, FUN=function(x) length(unique(x)) )