Create new column with values from certain rows of other columns - r

Question

I have a dataframe that looks like this:
(Example edited)
df <- data.frame(Subject = c(rep("A", 9), rep("B", 8)),
Trial = c(1,1,2,3,4,4,5,6,6,1,2,2,3,4,5,5,6),
Feature_1 = c(rep(123, 2), 234, 345, rep(456, 2), 567, rep(678, 2), 831, rep(444, 2), 461, 921, rep(436, 2), 111),
Feature_2 = c(rep(321, 2), 543, 654, rep(765, 2), 876, rep(987, 2), 912, rep(302, 2), 900, 555, rep(382, 2), 197),
Feature_3 = c(rep(190, 2), 459, 392, rep(398, 2), 492, rep(587, 2), 761, rep(901, 2), 783, 312, rep(880, 2), 229),
Feature_correct = NA)
df
Subject Trial Feature_1 Feature_2 Feature_3 Feature_correct
1 A 1 123 321 190 NA
2 A 1 123 321 190 NA
3 A 2 234 543 459 NA
4 A 3 345 654 392 NA
5 A 4 456 765 398 NA
6 A 4 456 765 398 NA
7 A 5 567 876 492 NA
8 A 6 678 987 587 NA
9 A 6 678 987 587 NA
10 B 1 831 912 761 NA
11 B 2 444 302 901 NA
12 B 2 444 302 901 NA
13 B 3 461 900 783 NA
14 B 4 921 555 312 NA
15 B 5 436 382 880 NA
16 B 5 436 382 880 NA
17 B 6 111 197 229 NA
What I need is for the Feature_correct column to contain the values from Feature_n depending on Trial for each Subject. So:
Subject A & Trials 1 and 2: Feature_correct contains the values for Subject A & Trials 1 and 2 under Feature_1 (respectively).
Subject A & Trials 3 and 4: Feature_correct contains the values for Subject A & Trials 3 and 4 under Feature_2 (respectively).
Subject A & Trials 5 and 6: Feature_correct contains the values for Subject A & Trials 5 and 6 under Feature_3 (respectively).
and so on for Subject B.
This is my goal:
df$Feature_goal <- c(rep(123, 2), 234, 654, rep(765, 2), 492, rep(587, 2), 831, rep(444, 2), 900, 555, rep(880, 2), 229)
head(df)
Subject Trial Feature_1 Feature_2 Feature_3 Feature_correct Feature_goal
1 A 1 123 321 190 NA 123
2 A 1 123 321 190 NA 123
3 A 2 234 543 459 NA 234
4 A 3 345 654 392 NA 654
5 A 4 456 765 398 NA 765
6 A 4 456 765 398 NA 765
I know how to do this manually (specifying the Subject name and Trial number in the syntax), but I'd like to create a loop (or whatever else works) so that I don't have to type the name of each Subject (in my real dataset I have many participants and many "Feature" variables).
I've tried this for loop, but I get an error:
df <- for(i in 1:nrow(df$Subject)) {
if(df$Trial %in% c(1,2)){
df[df$Subject == i $ df$Trial %in% c(1,2),]$Feature_correct = df[df$Subject == i & df$Trial %in% c(1,2),]$Feature_1
}
if(df$Trial %in% c(3,4)){
df[df$Subject == i $ df$Trial %in% c(3,4),]$Feature_correct = df[df$Subject == i & df$Trial %in% c(3,4),]$Feature_2
}
if(df$Trial %in% c(5,6)){
df[df$Subject == i $ df$Trial %in% c(5,6),]$Feature_correct = df[df$Subject == i & df$Trial %in% c(5,6),]$Feature_3
}
}
> Error in 1:nrow(df$Subject) : argument of length 0
Indeed,
nrow(df$Subject)
> NULL
Would anyone know how to make this work (with the loop or in any other way)?

Answer 1

A vectorized way would be to create a row/column index by pasting "Feature" with Trial number to match it with column names and subset values from original dataframe.
df$Feature_Goal <- df[cbind(seq_len(nrow(df)),
match(paste0("Feature_", df$Trial), names(df)))]
df
# Subject Trial Feature_1 Feature_2 Feature_3 Feature_correct Feature_Goal
#1 A 1 123 321 190 NA 123
#2 A 1 123 321 190 NA 123
#3 A 2 234 543 459 NA 543
#4 A 2 234 543 459 NA 543
#5 A 3 345 654 392 NA 392
#6 A 3 345 654 392 NA 392
#7 B 1 456 765 398 NA 456
#8 B 1 456 765 398 NA 456
#9 B 2 567 876 492 NA 876
#10 B 2 567 876 492 NA 876
#11 B 3 678 987 587 NA 587
#12 B 3 678 987 587 NA 587

Answer 2

Here is a solution using a loop.
for (i in 1:3) {
idx <- which(df$Trial == i)
df[idx,6] <- df[idx,i+2]
}