dplyr mutate_if with multiple conditions?

Question

I have data that looks as the following:

dt <- data.frame(id = sample(LETTERS, 100) ,A = seq(1:100), B = sample(1:100000, 100), C = rpois(100,1), D = seq(1:100)^3)

I am trying to transform any variable that has a max value > 1000

dt %>% mutate_if(max(.) > 1000, log10)

For model fitting. But I keep getting the following error:

Error in tbl_if_vars(.tbl, .p, .env, ..., .include_group_vars = .include_group_vars) : 
  length(.p) == length(tibble_vars) is not TRUE

Please help!

Ronak Shah · Accepted Answer · 2019-12-10T03:17:42.463

1

Use a formula argument

library(dplyr)
dt %>% mutate_if(~is.numeric(.) && max(., na.rm = TRUE) > 1000, log10)

To make it more clear you can use an anonymous function call

dt %>% mutate_if(function(x) is.numeric(x) && max(x, na.rm = TRUE) > 1000, log10)

edited Dec 10 '19 at 03:17

answered Dec 10 '19 at 02:21

Ronak Shah

It's not working. I keep getting the following error: "Error: Only strings can be converted to symbols" – akash87 Dec 10 '19 at 02:30
Also, I added another column – akash87 Dec 10 '19 at 02:32
@akash87 To make it work only for numeric columns you can do `dt %>% mutate_if(~is.numeric(.) && max(.) > 1000, log10)` However, I don't get the error message that you receive. Are you using the same dataset shared in the post. – Ronak Shah Dec 10 '19 at 02:37
Still getting the "Error: Only strings can be converted to symbols" – akash87 Dec 10 '19 at 02:58
@akash87 are you using the same dataset as shared in the post? – Ronak Shah Dec 10 '19 at 02:59
No I am not, I am using a simulated dataset – akash87 Dec 10 '19 at 02:59
Does the above answer work on the data you have shared in the post? – Ronak Shah Dec 10 '19 at 02:59
Let us [continue this discussion in chat](https://chat.stackoverflow.com/rooms/203944/discussion-between-akash87-and-ronak-shah). – akash87 Dec 10 '19 at 03:00

1 Answers1