R filtering for strings across several columns-CodePudding

I have a dataset of several rows of characters. For reasons I can't get into here, I need to leave these columns as character columns, but need to filter out rows containing letters.

In short, something like this:

A	B	C
1	A	4
2	3	B
5	6	7

Without converting the columns into numeric columns, I need to filter out the first two rows because they contain text characters.

I tried the following but no luck:

df %>%
filter(str_detect( c(A:C), "[AZ]") != "[AZ]")

CodePudding user response：

library(dplyr)

df %>% 
  filter(if_all(.fns = function(x) as.numeric(x)))

# A tibble: 1 x 3
    A     B     C    
  <int> <chr> <chr>
1   5     6     7

CodePudding user response：

If you want to use stringr and str_detect you can try:

library(stringr)
library(dplyr)

df %>% 
  filter(across(A:C, ~!str_detect(., "[A-Z]")))

Or to filter based on all columns in the data.frame:

df %>% 
  filter(across(everything(), ~!str_detect(., "[A-Z]")))

Output

  A B C
1 5 6 7

CodePudding user response：

In base R you might do df[complete.cases(sapply(df, match, 0:9)),]