[R] Aggregating frequency of irregular time series

[Whit Armstrong]

This function will probably work for you if your dates are in the format
yyyymmdd.  Use it to convert the series from daily to monthly, then use a 1
period difference to calc the returns of the series.

Feel free to contact me off list with any questions.


Dates <- function(x) rownames(x)

Timebase <- function(x,monthly=T,first.day=T) {
	if(length(dim(x))!=2) stop("can only take 2 dimensional data.")

	seriesDates <- as.integer(Dates(x))
	
	# year and Month
	seriesYM <- as.integer(seriesDates/100)
	seriesYM.unique <- sort(unique(seriesYM))
	ans.nrow <- length(seriesYM.unique)

	# loop through all the unique months and grab the max day of month
	ans <- matrix(NA,nrow=ans.nrow,ncol=ncol(x))
	ans.dates <- vector("numeric",ans.nrow)
	
	for (i in 1:ans.nrow) {
	
		this.month <- seriesYM == seriesYM.unique[i]
		max.day <- max(seriesDates[this.month])		
		ans.dates[i] <- max.day
	}

	
	colnames(ans) <- colnames(x)
	ans[] <- x[as.character(ans.dates),]
	
	if(first.day==T) {
		rownames(ans) <-
paste(substring(as.character(ans.dates),1,6),"01",sep="")
	} else {
		rownames(ans) <- ans.dates
	}
	
	as.tseries(ans)
}

-----Original Message-----
From: r-help-bounces at stat.math.ethz.ch
[mailto:r-help-bounces at stat.math.ethz.ch] On Behalf Of Patrick Burns
Sent: Tuesday, March 30, 2004 2:37 PM
To: Ajay Shah
Cc: r-help at stat.math.ethz.ch; Ivan Alves
Subject: Re: [R] Aggregating frequency of irregular time series


Assuming that you are using log returns, one approach (probably not the
best) is to convert the dates to julian days, find all of the Sundays
spanning the dates in the series, then do something along the lines of:

tapply(the.log.returns, cut(julian.days, sundays), sum)


Patrick Burns

Burns Statistics
patrick at burns-stat.com
+44 (0)20 8525 0696
http://www.burns-stat.com
(home of S Poetry and "A Guide for the Unwilling S User")

Ajay Shah wrote:

>>S-Plus has the function AggregateSeries() whose name is self 
>>explanatory.  For instance one can derive monthly series from daily 
>>ones by specifying end-of-period, averages, sums, etc.  I looked for a 
>>similar function in the packages "its" and "tseries", but found 
>>nothing.  I also help.searched() for aggregate to no avail.  Would 
>>anybody be so kind to point me in the right direction?
>>    
>>
>
>I once needed a function which would convert daily stock prices into 
>weekly returns. I know, the code is pretty bad (it is all loops and 
>it's very slow), but my knowledge of R is weak and I really needed it, 
>so I just used brute force. See EOF for the function. Gabor and Dirk 
>and Brian Ripley and others on the list were very helpful to me in 
>getting to the point where I could write this, though obviously they 
>should not be blamed for my bad code! :-)
>
>I would be very happy if listers could give me ideas on how to do this 
>better.
>
>Daily to monthly is innately easier since months are 'more normal' than 
>weeks. I have perl code which does this, which supports 2 cases: 
>Reporting the last traded price (LTP) of the month versus reporting the 
>average of the month. If this is useful to you, ask me.
>
>What are the specialised finance libraries available with S-Plus? Can 
>one marry R with commercial S libraries? I don't like having a 
>dependence on commercial code, but I might be willing to compromise and 
>buy libraries that work with R.
>
>Thanks,
>
>  
>

______________________________________________
R-help at stat.math.ethz.ch mailing list
https://www.stat.math.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide!
http://www.R-project.org/posting-guide.html