Filter multiple values on a string column in dplyr

Tom O picture Tom O · Sep 3, 2014 · Viewed 213.9k times · Source

I have a data.frame with character data in one of the columns. I would like to filter multiple options in the data.frame from the same column. Is there an easy way to do this that I'm missing?

Example: data.frame name = dat

days      name
88        Lynn
11          Tom
2           Chris
5           Lisa
22        Kyla
1          Tom
222      Lynn
2         Lynn

I'd like to filter out Tom and Lynn for example.
When I do:

target <- c("Tom", "Lynn")
filt <- filter(dat, name == target)

I get this error:

longer object length is not a multiple of shorter object length

Answer

BrodieG picture BrodieG · Sep 3, 2014

You need %in% instead of ==:

library(dplyr)
target <- c("Tom", "Lynn")
filter(dat, name %in% target)  # equivalently, dat %>% filter(name %in% target)

Produces

  days name
1   88 Lynn
2   11  Tom
3    1  Tom
4  222 Lynn
5    2 Lynn

To understand why, consider what happens here:

dat$name == target
# [1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE  TRUE

Basically, we're recycling the two length target vector four times to match the length of dat$name. In other words, we are doing:

 Lynn == Tom
  Tom == Lynn
Chris == Tom
 Lisa == Lynn
 ... continue repeating Tom and Lynn until end of data frame

In this case we don't get an error because I suspect your data frame actually has a different number of rows that don't allow recycling, but the sample you provide does (8 rows). If the sample had had an odd number of rows I would have gotten the same error as you. But even when recycling works, this is clearly not what you want. Basically, the statement dat$name == target is equivalent to saying:

return TRUE for every odd value that is equal to "Tom" or every even value that is equal to "Lynn".

It so happens that the last value in your sample data frame is even and equal to "Lynn", hence the one TRUE above.

To contrast, dat$name %in% target says:

for each value in dat$name, check that it exists in target.

Very different. Here is the result:

[1]  TRUE  TRUE FALSE FALSE FALSE  TRUE  TRUE  TRUE

Note your problem has nothing to do with dplyr, just the mis-use of ==.