remove rows from dataframe based on value, ignoring NAs - r

Question

I have a dataframe that I would like to delete rows from, based on a value in a specific column.
As an example, the dataframe appears something like this:
a b c d
1 1 2 3 0
2 4 NA 1 NA
3 6 4 0 1
4 NA 5 0 0
I would like to remove all rows with a value greater than 0 in column d. I have been trying to use the following code to do this:
df <- df[!df$d > 0, ]
but this is appearing to have the effect of deleting all the value is rows with an NA value in column d. I was assuming that a na.rm = TRUEargument was needed but I wasn't sure where to fit it in the function above.
Cheers,
Ant

Answer 1

We need to select the rows where d is not greater than 0 OR there is NA in d
df[with(df, !d > 0 | is.na(d)), ]
# a b c d
#1 1 2 3 0
#2 4 NA 1 NA
#4 NA 5 0 0
Or we can also use subset
subset(df, !d > 0 | is.na(d))
or dplyr filter
library(dplyr)
df %>% filter(!d > 0 | is.na(d))
The !d > 0 part can also be reversed to
subset(df, d < 1 | is.na(d))
to get the same result.

Answer 2

We can construct the logical vector with complete.cases
subset(df, !d > 0 | complete.cases(d))
# a b c d
#1 1 2 3 0
#3 6 4 0 1
#4 NA 5 0 0
Or use subset with replace
subset(df, !replace(d, is.na(d), 0) > 0)
Or with tidyverse
library(tidyverse)
df %>%
filter(!replace_na(d, 0) >0)
which is slightly different from the method mentioned here or here
data
df <- structure(list(a = c(1L, 4L, 6L, NA), b = c(2L, NA, 4L, 5L),
c = c(3L, 1L, 0L, 0L), d = c(0L, NA, 1L, 0L)), class = "data.frame",
row.names = c("1", "2", "3", "4"))

Answer 3

If u add a |all rows that has a NA will match. The condition !df$d > 0 will get executed for those in d that are not a NA. So I think you were looking for:
df[is.na(df$d) | !df$d > 0, ]
Wheras, the below wont include the rows that has a NA in column d and that does not match the condition !df$d > 0
df[!is.na(df$d) & !df$d > 0, ]