# [R] Function with multiple indices

cfchang3 cfchang3 at nccu.edu.tw
Tue Apr 17 16:20:12 CEST 2012

```Hi,

I wrote a function which works for one group but not for multiple groups in several years. Could anyone help me out?

The dateset has 3 variables, year, group, and sales. I want to calculate the annual group median adjusted sales performance for observation i in group j and year yr, I do the following procedure.
For each yar,
1. exclude observation i from group j
2. calculate the sales median for group j without observation i and call it "med"
3. substract "med" from the sales of observation i to obtain the group median adjusted sales performance for observation i

The dataset, function, and questions are as follows.

1997  11 0.027
1997  11 0.001
1997  11 0.697
1997  11 0.047
1997  11 0.225
1998  11 0.025
1998  11 0.002
1998  11 0.001
1998  11 0.659
1998  11 0.037
1997  12 0.152
1997  12 0.025
1997  12 0.417
1997  12 0.081
1997  12 0.194
1997  12 0.069
1997  12 0.062
1998  12 0.146
1998  12 0.028
1998  12 0.437
1997  13 0.087
1997  13 0.008
1997  13 0.014
1997  13 0.010
1997  13 0.018
1997  13 0.002
1997  13 0.010
1997  13 0.079
1997  13 0.092
1997  13 0.118
1997  14 0.111
1997  14 0.013
1997  14 0.506
1997  14 0.333
1997  14 0.018
1997  14 0.017
1998  14 0.155
1998  14 0.018
1998  14 0.477

# The whole dataset
x1

# Data for the first group only
GROUP11 <- x1[1:10,]
GROUP11

# The function works for the first group in one year
perf = function(x) {
for (i in 1:nrow(x)) {
salesi <- x\$SALES[i]
med <- median(x\$SALES[-i])
print  (salesi - med)
}
}
perf(GROUP11)

# Why the above funtion doesn't work in the following?
tapply(x1\$SALES, list(x1\$YEAR, x1\$GROUP), perf)
aggregate(x1\$SALES, by=list(x1\$YEAR, x1\$GROUP),FUN=perf)

Thanks!

Chingfu Chang

```