If variable has same name as dataframe's colum, dplyr filter's not working

1
votes

I have a simple loop that iterate across a number of string values in a vector called measurements:

measurements <- c("A","B","C","D")

here a reproducible data frame:

value <- c(1,2,3,4)
measurement <- c("A","B","C","D")
questiondata <- data.frame(measurement, value)
questiondata <- as.tibble(questiondata)

At first, the loop filters rows based on the measurement column. If the variable assigned in the loop has the same name as the column name of my data frame the filter does not work, it prints the entire dataframe 4 times:

for (measurement in measurements){
  print(measurement)
  print(questiondata %>% dplyr::filter(measurement == measurement))
}

If, instead,I change the variable name - from "measurement" to "m" for instance- it works:

for (m in measurements){
  print(m)
  print(questiondata %>% dplyr::filter(measurement == m))
}

Does anyone know the reason of this behaviour?

2
rfor-loopdplyrfilter
I think it could be a simple typo: measurement in measurement should work (since in your reproducible code you use measurement and not measurements), even though you should consider a different naming ;) - Julian
@Julian, thanks. Actually it should be measurement in measurements as the vector measurements is assigned before. - spleen
@Darren Tsai, There is not typo. the vector "measurements" is assigned at the beginning and contains the "letters" I´m interested to filter out. I run the code in a new session: no errors - spleen

2 Answers

2
votes

This issue results from the ambiguity between data-variables and env-variables for data-masked functions like filter().

In the following code, the both measurement refer to the measurement column from the questiondata data, and hence there are no rows being filtered out.

questiondata %>% filter(measurement == measurement)

# # A tibble: 4 × 2
#   measurement value
#   <chr>       <dbl>
# 1 A               1
# 2 B               2
# 3 C               3
# 4 D               4

You could use the .env pronoun to make it explicit where to find objects.

questiondata %>% filter(measurement == .env$measurement)

# # A tibble: 1 × 2
#   measurement value
#   <chr>       <dbl>
# 1 D               4
0
votes

I am not sure if this is what you look for. Here is an example.

A="ABC"
print(A)
for (A in c("1", "2")) print(A)
print(A)

Then we will get

[1] "ABC"
[1] "1"
[1] "2"
[1] "2"

The value of A is replaced by the do-loop.