continuous subgroups with ddply

I would like to summarize my experimental data every time a condition changes.

For example:

> df=data.frame(tos=1:9, temp=rep(c(25,50,25), each=3), response=c(3.2,3.3,3.3, 6.5, 6.5, 6.5, 3.5,3.6,3.5))
> df
    time temp response
1   1   25      3.2
2   2   25      3.3
3   3   25      3.3
4   4   50      6.5
5   5   50      6.5
6   6   50      6.5
7   7   25      3.5
8   8   25      3.6
9   9   25      3.5

I would like to summarize this in this way:

temp response.mean
25      3.3
50      6.5
25      3.5

If use ddply like this:

library(plyr)

ddply(df, c("temp"), summarize, reponse.mean=mean(response)

the output is:

  temp response.mean
1   25           3.4
2   50           6.5

Is there a way 开发者_JAVA百科to accomplish this?

Here is one way to accomplish this

# find how many observations in each experiment
tmp1    = rle(df$temp)$lengths

# create a column referring to experiment number
df$expt = rep(1:length(tmp1), tmp1)

# compute means for each combination of temp and expt
ddply(df, .(expt, temp), summarize, response.mean = mean(response))

This produces the output

   expt temp response.mean
1    1   25      3.266667
2    2   50      6.500000
3    3   25      3.533333