R: Calculating average by team

Question

I have soccer results data in the following format (thousands of observations):

     Div  date       value     pts
1    E0 2011-08-13   Blackburn 0.0
2    E0 2011-08-13      Fulham 0.5
3    E0 2011-08-13   Liverpool 0.5
4    E0 2011-08-13   Newcastle 0.5
5    E0 2011-08-13         QPR 0.0
6    E0 2011-08-13       Wigan 0.5
7    E0 2011-08-14       Stoke 0.5
8    E0 2011-08-14   West Brom 0.0
9    E0 2011-08-15    Man City 1.0
10   E0 2011-08-20     Arsenal 0.0
11   E0 2011-08-20 Aston Villa 1.0

plus other variables. "value" is the team, pts is the final result (win/loss/draw) as a numerical value. I'm trying to add a new variable which is the average of this value over the last X games for the team in that row. How do I do this without using some horrible loop?

Answer 1

take a look at this

using the zoo package and rollmean and the plyr package's ddply:

library(zoo)
library(plyr)
dat <- data.frame(value=letters[1:5], pts=sample(c(0, 0.5, 1), 50, replace=T))
ddply(dat, .(value), summarise, rollmean(pts, k=5, align='right'))

however, as far as I understand a "rolling average" it shortens your data set by definition. you can supply a fill argument though:

ddply(dat, .(value), summarise, rollmean(pts, k=5, fill=NA, align='right'))

Answer 2

Try ave function from stats.

Trt <- gl(n=2, k=3, length=2*3, labels =c("A", "B"))
Y <- 1:6
Data <- data.frame(Trt, Y)
 Data
  Trt Y
1   A 1
2   A 2
3   A 3
4   B 4
5   B 5
6   B 6
Data$TrtMean <- ave(Y, Trt, FUN=mean)
Data
  Trt Y TrtMean
1   A 1       2
2   A 2       2
3   A 3       2
4   B 4       5
5   B 5       5
6   B 6       5