R DPLYR Count Values BY Group-CodePudding

HAVE = data.frame(  STUDENT =c( 1,1,2,2,2,3,3   ),
    TEST    =c( 'A','B','A','B','C','A','C' ))

WANT = data.frame(STUDENT=c(1,2,3),
A=c(1,1,1),
B=c(1,1,0),
C=c(0,1,1),
TOT=c(2,3,2),
TOT.NOT.A=c(1,2,1))

I have a vertical data and wish to convert to a horizontal data as shown above. I can do

WANT = HAVE %>% group_by(STUDENT) %>% mutate(TOT = n_distinct (TEST))

to get 'TOT' but I do not know how to get 'A' 'B' 'C' or 'TOT.NOT.A'

CodePudding user response：

We could reshape to 'wide' format with pivot_wider and get the "TOT"al columns

library(dplyr)
library(tidyr)
HAVE %>% 
  pivot_wider(names_from = TEST, values_from = TEST,
     values_fn = length, values_fill = 0) %>% 
  mutate(TOT = rowSums(across(-STUDENT), na.rm = TRUE),
   TOT_NOT_A = rowSums(across(B:C), na.rm = TRUE))

-output

# A tibble: 3 × 6
  STUDENT     A     B     C   TOT TOT_NOT_A
    <dbl> <int> <int> <int> <dbl>     <dbl>
1       1     1     1     0     2         1
2       2     1     1     1     3         2
3       3     1     0     1     2         1

Or using base R

out <- addmargins(table(HAVE), 2)
cbind(out, TOT_NOT_A = rowSums(out[, c("B", "C")]))
  A B C Sum TOT_NOT_A
1 1 1 0   2         1
2 1 1 1   3         2
3 1 0 1   2         1

CodePudding user response：

Here is an alternative approach combining rowwise with c_across and sum:

library(dplyr)

HAVE %>% 
  add_count(STUDENT, TEST) %>% 
  pivot_wider(names_from = TEST, values_from =n, values_fill = 0 ) %>% 
  rowwise() %>% 
  mutate(TOT = sum(c_across(A:C), na.rm = TRUE),
         TOT_NOT_A = sum(c_across(B:C), na.rm = TRUE))

 STUDENT     A     B     C   TOT TOT_NOT_A
    <dbl> <int> <int> <int> <int>     <int>
1       1     1     1     0     2         1
2       2     1     1     1     3         2
3       3     1     0     1     2         1