How to use the apply function in a pipe operator?

Question

Here are my data.

> df
     VAR1 VAR2 VAR3 VAR4 VAR5 VAR6 VAR7 VAR8  Group
ACSK   16    4    0   25    0    0    7   16 Group3
CKPD    9   19    0   17    0   11    6   13 Group3
AEXF   10    0    0   36    0    0    0    0 Group1
AGNS   34    0   77   30    0    0    0    0 Group3
AGRL   19    0    0   52    0    0    0    0 Group2
AJEH    5    0   40    8    0    0    0   13 Group3
AOKS   34    0   47   43    0    0    0    0 Group2
AYKR   14    0  186   18    0    0    0    0 Group2
MIEY   20  226    0  102    0    0  201    0 Group1
BLHR   13    0    0   50    0    0    0    0 Group1

I want to sum the number by row grouped by the Group variable.

     VAR1 VAR2 VAR3 VAR4 VAR5 VAR6 VAR7 VAR8  Group
   1   43  226    0            ...  201    0 Group1
   2   67    0  117            ...    0    0 Group2
   3   64   23  233            ...   13   42 Group3

I tried the code below but it did not work.

df %>% 
  group_by(Group) %>% 
  apply(., 1, sum) %>% 
  rbind(.)

data

structure(list(VAR1 = c(16L, 9L, 10L, 34L, 19L, 5L, 34L, 14L, 
20L, 13L), VAR2 = c(4L, 19L, 0L, 0L, 0L, 0L, 0L, 0L, 226L, 0L
), VAR3 = c(0, 0, 0, 77, 0, 40, 47, 186, 0, 0), VAR4 = c(25L, 
17L, 36L, 30L, 52L, 8L, 43L, 18L, 102L, 50L), VAR5 = c(0, 0, 
0, 0, 0, 0, 0, 0, 0, 0), VAR6 = c(0, 11, 0, 0, 0, 0, 0, 0, 0, 
0), VAR7 = c(7L, 6L, 0L, 0L, 0L, 0L, 0L, 0L, 201L, 0L), VAR8 = c(16L, 
13L, 0L, 0L, 0L, 13L, 0L, 0L, 0L, 0L), Group = c("Group3", "Group3", 
"Group1", "Group3", "Group2", "Group3", "Group2", "Group2", "Group1", 
"Group1")), row.names = c("ACSK", "CKPD", "AEXF", "AGNS", "AGRL", 
"AJEH", "AOKS", "AYKR", "MIEY", "BLHR"), class = "data.frame")

Have a look here: https://stackoverflow.com/a/32734549/5785085 — CIAndrews, Dec 02 '21 at 12:56

Ronak Shah · Accepted Answer · 2021-12-02T14:01:24.733

I am not sure what is your expected output but as a general advice I would suggest not to mix base R and tidyverse approaches.

If you want to calculate rowwise sum then you don't need to group_by at all.

library(dplyr)

df %>%
  mutate(result = rowSums(select(., starts_with('VAR'))))

#     VAR1 VAR2 VAR3 VAR4 VAR5 VAR6 VAR7 VAR8  Group result
#ACSK   16    4    0   25    0    0    7   16 Group3     68
#CKPD    9   19    0   17    0   11    6   13 Group3     75
#AEXF   10    0    0   36    0    0    0    0 Group1     46
#AGNS   34    0   77   30    0    0    0    0 Group3    141
#AGRL   19    0    0   52    0    0    0    0 Group2     71
#AJEH    5    0   40    8    0    0    0   13 Group3     66
#AOKS   34    0   47   43    0    0    0    0 Group2    124
#AYKR   14    0  186   18    0    0    0    0 Group2    218
#MIEY   20  226    0  102    0    0  201    0 Group1    549
#BLHR   13    0    0   50    0    0    0    0 Group1     63

Further, if you want to sum the result by Group you may further do -

df %>%
  mutate(result = rowSums(select(., starts_with('VAR')))) %>%
  group_by(Group) %>%
  summarise(result = sum(result))

#  Group  result
#  <chr>   <dbl>
#1 Group1    658
#2 Group2    413
#3 Group3    350

For the updated question you may do -

df %>%
  group_by(Group) %>%
  summarise(across(starts_with('VAR'), sum))

#   Group VAR1 VAR2 VAR3 VAR4 VAR5 VAR6 VAR7 VAR8
#1 Group1   43  226    0  188    0    0  201    0
#2 Group2   67    0  233  113    0    0    0    0
#3 Group3   64   23  117   80    0   11   13   42

Or in base R -

aggregate(.~Group, df, sum)

Thank you. I have edited my question and added the expected output. I want to calculate the colsums of each group. — Lin Caijin, Dec 02 '21 at 13:58
Thank you. So can the apply function be used in a pipe operator? — Lin Caijin, Dec 02 '21 at 14:03
Yes, it can be used. Eg - `df[-9] %>% apply(1, sum)` but for what you are doing (group operations) `apply` is not the right function to use. — Ronak Shah, Dec 02 '21 at 14:05

Peter K · Answer 2 · 2021-12-02T13:38:56.653

To sum within groups and across rows:

# Load packages
library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union

# Generate data
df <- data.frame(VAR1 = sample(x = 0:20, size = 12),
                 VAR2 = sample(x = 0:20, size = 12),
                 VAR3 = sample(x = 0:20, size = 12),
                 Group = rep(c('Group 1', 'Group 2', 'Group 3'), 4))

# Sum
df %>% 
    # Sum each column by group
    group_by(Group) %>% 
    summarise(across(.cols = where(is.numeric), ~sum(.x))) %>% 
    # Sum each row
    rowwise() %>% 
    mutate(Total = sum(c_across(cols = where(is.numeric)))) %>%
    ungroup()
#> # A tibble: 3 × 5
#>   Group    VAR1  VAR2  VAR3 Total
#>   <chr>   <int> <int> <int> <int>
#> 1 Group 1    26    39    40   105
#> 2 Group 2    51    41    43   135
#> 3 Group 3    39    30    42   111

^{Created on 2021-12-02 by the reprex package (v1.0.0)}

How to use the apply function in a pipe operator?

2 Answers2