Calculating %changes with the By()

Question

I am an inexperienced R user and have been struggling with the By() function and would appreciate your help. The task is simple, I have a longitudinal dataset (How do I decl

Answer 1

You can use a combination of plyr to handle the group by operation on ID and quantmod has a function for the percentage change named Delt.

require(plyr)
require(quantmod)

> ddply(dat, "ID", transform,  DeltaCol = Delt(Temp))
    ID     Date      Temp    X.Change Delt.1.arithmetic
1  AAA 1/1/2003 0.7498817          NA                NA
2  AAA 1/2/2003 0.6666616 -0.11097769     -0.1109776868
3  AAA 1/3/2003 0.7730799  0.15962876      0.1596287574
4  AAA 1/4/2003 0.6290236 -0.18634075     -0.1863407501
5  AAA 1/5/2003 0.7333124  0.16579462      0.1657946178
6  BBB 1/1/2003 0.7073398          NA                NA
7  BBB 1/2/2003 0.7649865  0.08149798      0.0814979813
8  BBB 1/3/2003 0.6622015 -0.13436192     -0.1343619242
9  BBB 1/4/2003 0.7744518  0.16951080      0.1695107963
10 BBB 1/5/2003 0.5082909 -0.34367645     -0.3436764522
11 CCC 1/1/2003 0.8368362          NA                NA
12 CCC 1/2/2003 0.8371368  0.00035922      0.0003592196
13 CCC 1/3/2003 0.8090166 -0.03359092     -0.0335909235
14 CCC 1/4/2003 0.6902775 -0.14676969     -0.1467696849
15 CCC 1/5/2003 0.7963571  0.15367669      0.1536766860

Alternatively, you can skip the plyr bit, calculate the Delta for the entire data.frame and then update the first row for each ID. There are lots of good ideas about selecting the first row of a data.frame based off of an identifier here. Something like this would probably work:

dat$Delta <- Delt(dat$Temp)
dat[ diff(c(0,dat$ID)) != 0, 5] <- NA

On a related note, if anyone can explain why Delta doesn't seem to accept my plea to give it a reasonable column name, I'd appreciate it.