subtracting a specific condition for each measure

Question

I have a data frame which looks like following and it continues up to subject 22 Beta is the dependent measure.

Subject ROI Block Condition Beta
1   motor1  1   nopred_noom -2.8653
1   motor1  1   pred_noom   -2.9126
1   motor1  1   nopred_om   -2.8688
1   motor1  1   pred_om -2.9098
1   motor1  1   null    -2.7717
1   motor1  2   nopred_noom -2.2382
1   motor1  2   pred_noom   -2.0583
1   motor1  2   nopred_om   -2.2207
1   motor1  2   pred_om -2.1928
1   motor1  2   null    -2.1166
1   motor1  3   nopred_noom -1.5992
1   motor1  3   pred_noom   -1.5493
1   motor1  3   nopred_om   -1.5230
1   motor1  3   pred_om -1.4851
1   motor1  3   null    -1.5624
2   motor1  1   nopred_noom -1.1354
2   motor1  1   pred_noom   -1.1614
2   motor1  1   nopred_om   -1.2779
2   motor1  1   pred_om -1.1234
2   motor1  1   null    -1.2203
2   motor1  2   nopred_noom -1.5728
2   motor1  2   pred_noom   -1.6614
2   motor1  2   nopred_om   -1.7076
2   motor1  2   pred_om -1.7702
2   motor1  2   null    -1.4170

There are 5 conditions, but I want to use the condition null as the baseline and want to subtract it from other conditions in each corresponding block and subject.

so I would subtract Beta in subject 1, block 1, condition "null" from Beta measures in other conditions in subject1, block 1 but then I want to use beta value "null" from subject1, block2 for measures in subject 1, block2 and so on.

null condition occurs every 5 conditions and i suspect I need to use a loop but I am quite new to R and I am not sure how to do this.

any help is appreciated!!! thanks :)

score 1 · Answer 1 · answered Jul 03 '13 at 12:01

This needs the so-called "split-apply-combine" approach. There are many possibilities to do this. The easiest for a beginner is package plyr, because of its nice syntax.

library(plyr)
DF <- ddply(DF, .(Subject,ROI,Block), transform, Beta0 = Beta-Beta[Condition=="null"])

#    Subject    ROI Block   Condition    Beta   Beta0
# 1        1 motor1     1 nopred_noom -2.8653 -0.0936
# 2        1 motor1     1   pred_noom -2.9126 -0.1409
# 3        1 motor1     1   nopred_om -2.8688 -0.0971
# 4        1 motor1     1     pred_om -2.9098 -0.1381
# 5        1 motor1     1        null -2.7717  0.0000
# 6        1 motor1     2 nopred_noom -2.2382 -0.1216
# 7        1 motor1     2   pred_noom -2.0583  0.0583
# 8        1 motor1     2   nopred_om -2.2207 -0.1041
# 9        1 motor1     2     pred_om -2.1928 -0.0762
# 10       1 motor1     2        null -2.1166  0.0000
# <snip>

score 1 · Answer 2 · answered Jan 14 '15 at 17:37

Here is base R code for the above task:

#-- split
dfs <- split(df, list(df$Block, df$Subject))
#-- apply
Beta0<-NULL
for (i in 1:length(dfs))
{Beta0 <- dfs[[i]]$Beta - dfs[[i]][dfs[[i]]$Condition=="null",]$Beta;
 dfs[[i]][,"Beta0"] <- Beta0}
#-- recombine
dfrc <- do.call(rbind, dfs)

df= original data frame; dfs = a list comprising all the split subgroups; dfrc = new data frame that should reproduce the results displayed above for the new column "Beta0".

I posted this because I had a similar data set with - analagously - a missing value of "Beta" for the condition "null" in one block. Plyr produced an error message "arguments imply differing number of rows: x, 0" and di dnot compute. The above code, however, produced NAs for that block but computed all the rest.

score 0 · Answer 3 · answered Jul 03 '13 at 12:12

Getting into data.table is fun too. Or maybe I'm just on a DT kick...

+1 for an answerable question with data.

df <- read.table(text ='Subject ROI Block Condition Beta
1   motor1  1   nopred_noom -2.8653
1   motor1  1   pred_noom   -2.9126
1   motor1  1   nopred_om   -2.8688
1   motor1  1   pred_om -2.9098
1   motor1  1   null    -2.7717
1   motor1  2   nopred_noom -2.2382
1   motor1  2   pred_noom   -2.0583
1   motor1  2   nopred_om   -2.2207
1   motor1  2   pred_om -2.1928
1   motor1  2   null    -2.1166
1   motor1  3   nopred_noom -1.5992
1   motor1  3   pred_noom   -1.5493
1   motor1  3   nopred_om   -1.5230
1   motor1  3   pred_om -1.4851
1   motor1  3   null    -1.5624
2   motor1  1   nopred_noom -1.1354
2   motor1  1   pred_noom   -1.1614
2   motor1  1   nopred_om   -1.2779
2   motor1  1   pred_om -1.1234
2   motor1  1   null    -1.2203
2   motor1  2   nopred_noom -1.5728
2   motor1  2   pred_noom   -1.6614
2   motor1  2   nopred_om   -1.7076
2   motor1  2   pred_om -1.7702
2   motor1  2   null    -1.4170', header=T)

dt <- data.table(df)

delta_maker <- function(x) {
  return(x - x[5])
}

dt[, delta := delta_maker(Beta), by = list(ROI, Subject, Block)]

#Subject    ROI Block   Condition    Beta   delta
#1:       1 motor1     1 nopred_noom -2.8653 -0.0936
#2:       1 motor1     1   pred_noom -2.9126 -0.1409
#3:       1 motor1     1   nopred_om -2.8688 -0.0971
#4:       1 motor1     1     pred_om -2.9098 -0.1381
#5:       1 motor1     1        null -2.7717  0.0000
#6:       1 motor1     2 nopred_noom -2.2382 -0.1216
#7:       1 motor1     2   pred_noom -2.0583  0.0583
#8:       1 motor1     2   nopred_om -2.2207 -0.1041
#9:       1 motor1     2     pred_om -2.1928 -0.0762
#10:       1 motor1     2        null -2.1166  0.0000
#11:       1 motor1     3 nopred_noom -1.5992 -0.0368
#12:       1 motor1     3   pred_noom -1.5493  0.0131
#13:       1 motor1     3   nopred_om -1.5230  0.0394
#14:       1 motor1     3     pred_om -1.4851  0.0773
#15:       1 motor1     3        null -1.5624  0.0000
#16:       2 motor1     1 nopred_noom -1.1354  0.0849
#17:       2 motor1     1   pred_noom -1.1614  0.0589
#18:       2 motor1     1   nopred_om -1.2779 -0.0576
#19:       2 motor1     1     pred_om -1.1234  0.0969
#20:       2 motor1     1        null -1.2203  0.0000
#21:       2 motor1     2 nopred_noom -1.5728 -0.1558
#22:       2 motor1     2   pred_noom -1.6614 -0.2444
#23:       2 motor1     2   nopred_om -1.7076 -0.2906
#24:       2 motor1     2     pred_om -1.7702 -0.3532
#25:       2 motor1     2        null -1.4170  0.0000
#Subject    ROI Block   Condition    Beta   delta

this uses the add-on package [datatable](http://cran.r-project.org/web/packages/data.table/index.html), which you will first need to install and then `library(datatable)`. — dirkjot, Jun 24 '14 at 11:05

subtracting a specific condition for each measure

3 Answers3

Linked