Error assigned data must be compatible with existing data-CodePudding

I want to create a new variable, "F", by adding columns (B C D E) if the column "A" is 1.

ID	A	B	C	D	E
001	1	1	2	NA	1
002	0	2	1	1	NA

df$F <- rowSums(df[df$A == '1', c(3:6)],na.rm=TRUE)

I get this error:

Error:
! Assigned data `rowSums(df[df$A == "1", c(3:6)], na.rm = TRUE)` must be compatible with existing data.
✖ Existing data has 12358 rows.
✖ Assigned data has 474 rows.
ℹ Only vectors of size 1 are recycled.
Backtrace:
  1. base::`$<-`(`*tmp*`, F, value = `<dbl>`)
 12. tibble (local) `<fn>`(`<vctrs___>`)
 Error:

How can I fix this? Are there other ways to get my final outcome something looks like the one below?

ID	A	B	C	D	E	F
001	1	1	2	NA	1	4
002	0	2	1	1	NA	NA

CodePudding user response：

Try this.

df$F <- ifelse(df$A %in% 1, rowSums(df[, c("B", "C", "D", "E")], na.rm=TRUE), NA)
df
#   ID A B C  D  E  F
# 1  1 1 1 2 NA  1  4
# 2  2 0 2 1  1 NA NA

Note, that due to using %in%, NAs in column A are treated like 0.

CodePudding user response：

We just need the logical to be on the lhs as well to keep the lengths same

df$F[df$A == '1'] <- rowSums(df[df$A == '1', c(3:6)],na.rm=TRUE)

-output

> df
  ID A B C  D  E  F
1  1 1 1 2 NA  1  4
2  2 0 2 1  1 NA NA

CodePudding user response：

A tidyverse approach:

Libraries

library(dplyr)

Data

data <-
tibble::tribble(
    ~ID, ~A, ~B, ~C, ~D, ~E,
  "001", 1L, 1L, 2L, NA, 1L,
  "002", 0L, 2L, 1L, 1L, NA
  )

Code

data %>% 
  rowwise() %>% 
  mutate(`F` = if_else(A == 1, sum(c_across(cols = B:E),na.rm = TRUE), NA_integer_) )

Output

# A tibble: 2 x 7
# Rowwise: 
  ID        A     B     C     D     E     F
  <chr> <int> <int> <int> <int> <int> <int>
1 001       1     1     2    NA     1     4
2 002       0     2     1     1    NA    NA