使用“any”函数跨多列的逻辑函数

Question

I would like to run a logical operation (multiple conditions) across many columns. I have written a query which is working fine. However, I want to shorten my code as I have to write several queries.

I have tried shortening the query using "any" and "brackets". However, the second query is running fine but giving me a different answer. Does "any" function work on multiple columns?

Here are my conditions -

1. any of the column (B2 to B5) has 1 & B1 <=2, then "Noissue"
2. any of the column (B2 to B5) has -99 & B1 <=2, then "Noissue"
3. B1 ==3, then "Noissue"
4. Rest is all issue

Participate	B1	B2	B3	B4	B5	Query1	Query2
3	-1	-1	-1	-1	-1	Noissue	Noissue
1	-1	1	-1	-1	1	Noissue	Noissue
1	-1	-1	-1	-1	-1	Issue	Noissue
2	-1	1	1	-1	1	Noissue	Noissue
2	1	1	1	1	-1	Noissue	Noissue
1	-99	-99	-99	-99	-99	Noissue	Noissue

I appreciate if anyone help me on reducing the code lines using different functions.

 mutate(Batch_v1, 
               case_when (
                 ((Batch_v1$B1 == 1 |  Batch_v1$B2 == 1 | Batch_v1$B3 == 1 | Batch_v1$B4 == 1 | Batch_v1$B5 == 1| Batch_v1$B6 == 1| Batch_v1$B7 == 1|Batch_v1$B8 == 1|Batch_v1$B9 == 1|Batch_v1$B10 == 1|Batch_v1$BOth == 1) & 
                    Batch_v1$Participate %in% c(1,2,-99))~"Noissue",
                 ((Batch_v1$B1 == -99 |  Batch_v1$B2 == -99 | Batch_v1$B3 == -99|Batch_v1$B4 == -99 |Batch_v1$B5 == -99|Batch_v1$B6 == -99|Batch_v1$B7 == -99|Batch_v1$B8 == 1|Batch_v1$B9 == -99|Batch_v1$B10 == -99|Batch_v1$BOth == -99) & 
                    Batch_v1$Participate %in% c(1,2,-99))~"Noissue",
                 Batch_v1$Participate ==3 ~ "Noissue",
                 TRUE ~ "Issue"))





mutate(Batch_v1, 
   case_when (
     ((any(Batch_v1[,2:6] == 1)) & Batch_v1$Participate %in% c(1,2,-99))~ "Noissue",
     ((any(Batch_v1[,2:6] == -99)) & Batch_v1$Participate %in% c(1,2,-99))~ "Noissue",
     Batch_v1$Participate ==3 ~ "Noissue",
     TRUE ~ "Issue"))

Answer 1

We could uses across with case_when

library(dplyr)
df %>% 
    mutate(across(B2:B5, ~case_when(. == 1 & B1 <=2 ~ "Noissue",
                                    . == -99 & B1 <=2 ~ "Noissue",
                                    B1 == 3 ~ "Noissue",
                                    TRUE ~ "issue")
                  )
           )

Output:

  Participate    B1 B2      B3      B4      B5      Query1  Query2 
        <dbl> <dbl> <chr>   <chr>   <chr>   <chr>   <chr>   <chr>  
1           3    -1 issue   issue   issue   issue   Noissue Noissue
2           1    -1 Noissue issue   issue   Noissue Noissue Noissue
3           1    -1 issue   issue   issue   issue   Issue   Noissue
4           2    -1 Noissue Noissue issue   Noissue Noissue Noissue
5           2     1 Noissue Noissue Noissue issue   Noissue Noissue
6           1   -99 Noissue Noissue Noissue Noissue Noissue Noissue

data:

df <- structure(list(Participate = c(3, 1, 1, 2, 2, 1), B1 = c(-1, 
-1, -1, -1, 1, -99), B2 = c(-1, 1, -1, 1, 1, -99), B3 = c(-1, 
-1, -1, 1, 1, -99), B4 = c(-1, -1, -1, -1, 1, -99), B5 = c(-1, 
1, -1, 1, -1, -99), Query1 = c("Noissue", "Noissue", "Issue", 
"Noissue", "Noissue", "Noissue"), Query2 = c("Noissue", "Noissue", 
"Noissue", "Noissue", "Noissue", "Noissue")), problems = structure(list(
row = 6L, col = "Query2", expected = "", actual = "embedded null", 
file = "'test'"), row.names = c(NA, -1L), class = c("tbl_df", 
"tbl", "data.frame")), class = c("spec_tbl_df", "tbl_df", "tbl", 
"data.frame"), row.names = c(NA, -6L))

Answer 2

Whenever we have to use logical conditions rowwise across many columns, two main approaches should usually be considered. These obviate the need for rowwise() and Reduce() in the alternative with lapply/map %>% Reduce/reduce , or complex case_when() statements.

-1) rowSums(condition)
-2) if_any() / if_all()

This question is most suited for a solution with if_any() .

With if_any()

Batch_v1 %>% mutate(query3 = ifelse(if_any(B2:B5, ~.x %in% c(-99, 1)) & B1<=2,
              "Noissue",
              "Issue"))

With rowSums()

Batch_v1 %>% mutate(query3 = ifelse(rowSums(across(B2:B5, ~.x %in% c(-99, 1)))>0 & B1<=2,
                              "Noissue",
                              "Issue"))

Output

# A tibble: 6 x 9
  Participate    B1    B2    B3    B4    B5 Query1  Query2  query3 
        <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <chr>   <chr>   <chr>  
1           3    -1    -1    -1    -1    -1 Noissue Noissue Issue  
2           1    -1     1    -1    -1     1 Noissue Noissue Noissue
3           1    -1    -1    -1    -1    -1 Issue   Noissue Issue  
4           2    -1     1     1    -1     1 Noissue Noissue Noissue
5           2     1     1     1     1    -1 Noissue Noissue Noissue
6           1   -99   -99   -99   -99   -99 Noissue Noissue Noissue

There are some good answers to similar questions in here:
Rowwise logical operations with mutate() and filter() in R and here:
R - Remove rows from dataframe that contain only zeros in numeric columns, base R and pipe-friendly methods?

Answer 3

You could use

library(dplyr)

Batch_v1 %>% 
  rowwise() %>%
  mutate(
    Query3 = case_when(
      any(B1:B5 == 1)   & Participate %in% c(1,2,-99) ~ "Noissue",
      any(B1:B5 == -99) & Participate %in% c(1,2,-99) ~ "Noissue",
      Participate == 3                                ~ "Noissue",
      TRUE                                            ~ "Issue"
      )
    )

which returns

# A tibble: 6 x 9
# Rowwise: 
  Participate    B1    B2    B3    B4    B5 Query1  Query2  Query3 
        <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <chr>   <chr>   <chr>  
1           3    -1    -1    -1    -1    -1 Noissue Noissue Noissue
2           1    -1     1    -1    -1     1 Noissue Noissue Noissue
3           1    -1    -1    -1    -1    -1 Issue   Noissue Issue  
4           2    -1     1     1    -1     1 Noissue Noissue Noissue
5           2     1     1     1     1    -1 Noissue Noissue Noissue
6           1   -99   -99   -99   -99   -99 Noissue Noissue Noissue

The main problem with your second code is the function

any(Batch_v1[,2:6] == 1)

Let's take a look at

Batch_v1[,2:6] == 1

#>         B1    B2    B3    B4    B5
#> [1,] FALSE FALSE FALSE FALSE FALSE
#> [2,] FALSE  TRUE FALSE FALSE  TRUE
#> [3,] FALSE FALSE FALSE FALSE FALSE
#> [4,] FALSE  TRUE  TRUE FALSE  TRUE
#> [5,]  TRUE  TRUE  TRUE  TRUE FALSE
#> [6,] FALSE FALSE FALSE FALSE FALSE

So Batch_v1[,2:6] == 1 returns a data.frame of booleans. Applying any on this data.frame returns TRUE if any of the values inside this data.frame is TRUE . That's clearly not your desired behaviour. Using rowwise() forces any to be applied... well... per row.

Note: Inside a tidyverse -pipe, you don't want to use Batch_v1$B1 if you are refering on the current object you are working with. Batch_v1$B1 for example refers to the original Batch_v1 , without any transformations done. In this case, there is no real difference, but you shouldn't rely on this in general.