CODEWITHSUNDEEP
rdplyr
Lloyd Christmas
Lloyd Christmas

Reputation: 1038

refer to specific values of columns in dplyr

This answer is out there somewhere, but I'm having a hard time articulate it.

I want to refer to a subset a vector with a dplyr framework. It's best explained with an example.

I'm trying to calculate something called "Delta E"

enter image description here

Which requires me to refer to one row (called "Control") in this example.

I've tried to find ways to refer to specific values within a column of a data.frame.

In R: subset or dplyr::filter with variable from vector

Here's the example data:

df <- read.table(text="
sampleID     L     a     b
control   54.1  32.5  85.6
A         51.7  34.2  83.3
B         53.7  32.8  85.0
D         53.8  32.7  85.1", header=T)

And I'd like to use dplyr to get the end result, without doing anything to hacky.

df %>%
   mutate(deltaE = sqrt((Lc - L)^2 - (ac - a)^2 + (bc - b)^2)

  sampleID     L     a     b deltaE
  control   54.1  32.5  85.6  0    
  A         51.7  34.2  83.3  3.73 
  B         53.7  32.8  85.0  0.733
  D         53.8  32.7  85.1  0.614

Where Lc = 54.1, ac = 32.5 and bc = 85.6 which correspond to the control sample.

Upvotes: 1

Views: 200

Answers (1)

MrFlick
MrFlick

Reputation: 206616

It's not super pretty, but you can do

df %>%
  mutate(deltaE = sqrt((L[sampleID=="control"] - L)^2 + 
                         (a[sampleID=="control"] - a)^2 +
                         (b[sampleID=="control"] - b)^2))

Or split it up into steps

df %>%
  mutate(isControl = sampleID=="control",
         Lc = L[isControl],
         ac = a[isControl],
         bc = b[isControl]) %>% 
  mutate(deltaE = sqrt((Lc - L)^2 + 
                         (ac - a)^2 +
                         (bc - b)^2)) %>% 
  select(-isControl, -Lc, -ac, -bc)

Upvotes: 2

Related Questions