[R] using ddply but preserving some of the outside data

[David Winsemius]

library(zoo)
?rollmean

On Aug 5, 2009, at 3:00 PM, Jarrett Byrnes wrote:

> I have a bit of a quandy.  I'm working with a data set for which I  
> have sampled sites at a variety of dates.  I want to use this data,  
> and get a running average of the sampled values for the current and  
> previous date.
>
> I originally thought something like ddply would be ideal for this,  
> however, I cannot break up my data by date, and then apply a  
> function that requires information about the previous dates.
>
> I had thought to use a for loop and merge, but that doesn't quite  
> seem to be working.
>
> So, my questions are twofold
>
> 1) Is there a way to use something like the plyr library to do this  
> efficiently
> 	1a) Indeed, is there a way to use ddply or its ilk to have a  
> function that returns a vector of values, and then assign the  
> variables you are sorting by to the whole vector?  Or maybe making  
> each value it's own column in the new data frame, and then using  
> reshape is the answer.  Hrm.  Seems clunky.
>
> 2) Or, can a for loop around a plyr-kind of statement do the trick  
> (and if so, pointers on why the below code won't work) (also, it,  
> too, seems clunkier than I would like)
>
>
> sites<-c("a", "b", "c")
> dates<-1:5
>
> a.df<-expand.grid(sites=sites, dates=dates)
> a.df$value<-runif(15,0,100)
> a.df<-as.data.frame(a.df)
>
>
> #now, I want to get the average of the
> mean2<-function(df, date){
> 	sub.df<-subset(df, df$dates-date<1 &
> 				df$dates-date>-1 )
> 	return(mean(df$value))
> 	}
>
> my.df<-data.frame(sites=NA, dates=NA, V1=NA)
> for(a.date in a.df$dates){
> 	new.df<-ddply(a.df, "sites", function(df) mean2 (df, a.date))
> 	my.df<-merge(my.df, new.df) #doesn't seem to work
> }
>
> my.df
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.

David Winsemius, MD
Heritage Laboratories
West Hartford, CT