Sum across multiple columns with pattern conditionally-CodePudding

I want to sum across multiple columns that have a particular pattern for the column name.

The following works:

sum = rowSums(across(matches('pattern')), na.rm = TRUE)

However, I want to only sum if the value is 1 or NA (0). So if the value is 2 for example, it will ignore it and essentially count it as a zero. Would the which() function help with this?

For example:

The total of 0,NA,1,1,1,0 would be 3

The total of 0,NA,2,3,1,NA would be 1

Thanks!

CodePudding user response：

For dummy data x defined like

x <- data.frame(
  v1 = c(0,NA,2,3,1,NA),
  v2 = c(0,NA,1,1,1,0)
  ) %>% t

   [,1] [,2] [,3] [,4] [,5] [,6]
v1    0   NA    2    3    1   NA
v2    0   NA    1    1    1    0

You may try rowSums(ifelse(x>1, 0, x), na.rm = T)

v1 v2 
 1  3

CodePudding user response：

If I have understood correctly, you only want to count number of 1's in columns with specific pattern which can be done with rowSums as -

library(dplyr)

df <- data.frame(a1 = c(1, 0, 2, NA, 1), a2 = c(1, 3, 0, 4, NA))

df %>%
  mutate(sum = rowSums(dplyr::select(., matches('a')) == 1, na.rm = TRUE))

#  a1 a2 sum
#1  1  1   2
#2  0  3   0
#3  2  0   0
#4 NA  4   0
#5  1 NA   1