CODEWITHSUNDEEP
rdplyr
Julia M
Julia M

Reputation: 133

Rounding all numeric columns but one in table

I have created numerous output tables from t-tests and ANOVAs and I'd like to round all numeric columns of the tables apart from the column containing the p-values (p.value).

Current code:

library(dplyr)
library(broom)

a <- rnorm(100, 0.75, 0.1)

t.test <- t.test(a, mu = 0.5, alternative = "greater") %>%
          broom::tidy() %>%
          mutate_if(is.numeric, round, 2)

The issue is that this also rounds my p-value which is then displayed as 0. I already have a function for reporting p-values for my markdown file so I'm wondering how I can keep the p-value (p.value) unchanged yet round all other numeric columns to 2 digits?

Thanks

Upvotes: 3

Views: 4437

Answers (3)

symbolrush
symbolrush

Reputation: 7457

If you really wanna round all numeric columns but the p.value column, just make shure your p.value column doesn't get rounded by coercing it to a character before the rounding and then back to numeric after rounding.

library(dplyr)
library(broom)

a <- rnorm(100, 0.75, 0.1)

t.test <- t.test(a, mu = 0.5, alternative = "greater") %>%
  broom::tidy() %>%
  mutate(p.value = as.character(p.value)) %>%
  mutate_if(is.numeric, round, 2) %>%
  mutate(p.value = as.numeric(p.value))
t.test
# A tibble: 1 x 8
  estimate statistic  p.value parameter conf.low conf.high method            alternative
     <dbl>     <dbl>    <dbl>     <dbl>    <dbl>     <dbl> <chr>             <chr>      
1     0.76      28.3 1.52e-49        99     0.74       Inf One Sample t-test greater

If you just want to round some columns, then you probably be better of using mutate_at as @user5249203 promotes.

Upvotes: 6

user5249203
user5249203

Reputation: 4648

You can do simple mutate_at

library(dplyr)
library(broom)

a <- rnorm(100, 0.75, 0.1)
t.test(a, mu = 0.5, alternative = "greater") %>%
  broom::tidy() %>% 
  mutate_at(vars(- c(p.value,method,alternative)), round, 2)

#> # A tibble: 1 x 8
#>   estimate statistic  p.value parameter conf.low conf.high method
#>      <dbl>     <dbl>    <dbl>     <dbl>    <dbl>     <dbl> <chr> 
#> 1     0.75      27.1 5.89e-48        99     0.74       Inf One S~
#> # ... with 1 more variable: alternative <chr>

Created on 2019-04-18 by the reprex package (v0.2.1)

Upvotes: 2

TinglTanglBob
TinglTanglBob

Reputation: 647

not very nice and not dplyr, but i think it should do the job :)

select all elements from t.test which are numeric and not p.value and apply round to them

t.test[(names(t.test)[which(!names(t.test) %in% c("p.value")  & sapply(t.test, class) == "numeric")])] = 
  lapply(t.test[(names(t.test)[which(!names(t.test) %in% c("p.value")  & sapply(t.test, class) == "numeric")])], round, 2)

Upvotes: 0

Related Questions