CODEWITHSUNDEEP
rdataframedplyrdcast
Santi
Santi

Reputation: 368

A special case of dcast in R

My question seems really simple, and indeed I feel very annoyed for the fact that I cannot make it work. Let's say I have a simple dataframe with one column for group and one variable x. Because my variable group contains a "control" condition, I would like to run a t.test of all my other conditions against my control variable. 

library(data.table) # I am use to the data.table sintax, tho I will happily accept a solution in any other dialect

# Generate dummy data
set.seed(1)
 df <- data.table(x = rnorm(100), g = sample(LETTERS[1:3], size = 100, replace =T ))
setkey(df, g, x) # Order

 df # Inspect data

For that purpose, I would like to dcast the control group and add it as a new column. Since what I want is to run a t-test and for it, I will use the whole group, I do not mind in which order the column gets included. However, the function that I would use to change from a long format to a wide format (dcast), doesn't seem to work here.

# dcast appoach
m <- dcast(df, x ~ g) # This is just... B*#!!it

So here is an approximation of what I look for:

# Kind of what I want

# Isolate control condition
Control <- df[g == "C"] 

df[, C := rep(Control, 3)] # In this case it says there a "remainder", tho I would prefer to add NAs to the variable x until completion

I also would not mind having all the groups A, B and C, as columns.

Thanks in advance for your help

Upvotes: 0

Views: 108

Answers (1)

Uwe
Uwe

Reputation: 42544

Perhaps, this might be what the OP has asked for:

library(data.table)
dcast(df, rowid(g) ~ g, value.var = "x")

     g            A           B           C
 1:  1 -1.804958629 -1.98935170 -2.21469989
 2:  2 -1.470752384 -1.52356680 -0.74327321
 3:  3 -1.276592208 -1.37705956 -0.62124058
 4:  4 -1.253633400 -1.12936310 -0.61202639
 5:  5 -1.224612615 -1.04413463 -0.58952095
 6:  6 -0.934097632 -0.83562861 -0.47340064
 7:  7 -0.709946431 -0.82046838 -0.41499456
 8:  8 -0.707495157 -0.68875569 -0.39428995
 9:  9 -0.626453811 -0.47815006 -0.30538839
10: 10 -0.573265414 -0.25336168 -0.13505460
11: 11 -0.568668733 -0.13517862  0.02800216
12: 12 -0.542520031 -0.11234621  0.39810588
13: 13 -0.443291873 -0.05931340  0.41794156
14: 14 -0.367221476 -0.05612874  0.55848643
15: 15 -0.304183924 -0.05380504  0.61982575
16: 16 -0.164523596 -0.01619026  0.69696338
17: 17 -0.155795507  0.07434132  0.82122120
18: 18 -0.102787727  0.15325334  0.88110773
19: 19 -0.044933609  0.34111969  0.94383621
20: 20 -0.039240003  0.36458196  1.12493092
21: 21  0.001105352  0.38767161  1.16040262
22: 22  0.074564983  0.48742905  1.17808700
23: 23  0.183643324  0.56971963  1.46555486
24: 24  0.188792300  0.59390132  1.51178117
25: 25  0.267098791  0.61072635          NA
26: 26  0.291446236  0.76317575          NA
27: 27  0.329507772  1.10002537          NA
28: 28  0.332950371  1.35867955          NA
29: 29  0.370018810  1.43302370          NA
30: 30  0.389843236  1.58683345          NA
31: 31  0.475509529  2.40161776          NA
32: 32  0.556663199          NA          NA
33: 33  0.575781352          NA          NA
34: 34  0.593946188          NA          NA
35: 35  0.689739362          NA          NA
36: 36  0.700213650          NA          NA
37: 37  0.738324705          NA          NA
38: 38  0.768532925          NA          NA
39: 39  0.782136301          NA          NA
40: 40  0.918977372          NA          NA
41: 41  1.063099837          NA          NA
42: 42  1.207867806          NA          NA
43: 43  1.595280802          NA          NA
44: 44  1.980399899          NA          NA
45: 45  2.172611670          NA          NA
     g            A           B           C

This works by artificially introducing an individual row count rowid(g) for each group.

However, in line with 42-'s comment, I do not understand how this will help to solve OP's underlying problem.

Upvotes: 1

Related Questions