[R] Writing output to a file in a loop

[Roger Bivand]

On Thu, 10 Feb 2005, Wojtek Slusarski wrote:

> Hello,
> My problem is, that I have to build hundreds of GARCH models to obtain 
> volatility forecasts. I would like to run a loop, that would build those 
> forecasts for me. There is no problem, with writing only the results of 
> the forecasts, but I'd like to have stored results of the models in some 
> file, that I could check later, what are the models like, to be able to 
> compare if I should use GARCH(1,1) or GARCH(0,4) or any other, that 
> suits the data best. The data file looks like that:
> 
> TICKER;DTYYYYMMDD;OPEN;HIGH;LOW;CLOSE;VOL
> WIG20;19940414;1000,00;1000,00;1000,00;1000,00;71600
> WIG20;19940418;1050,47;1050,47;1050,47;1050,47;99950
> WIG20;19940419;1124,92;1124,92;1124,92;1124,92;138059
> ...
> 
> I wrote a script to do that, but when I use sink() to save the results 
> in a text file, after running the script files are empty. I read in FAQ, 
> that in a loop R only computes and prints only some warning messages, 
> but I don't know how to do that in any other way, to ommit this problem.
> Below I enclose the code:
> 
> ############################################################
> 
> library(tseries)
> 
> wig20 <- read.csv("wig20.txt", sep=";", dec=",")
> m <- 2321   	#upper bound of time series
> niter <- 10        #length(wig20$CLOSE)- m
> fcv <- 0
> 
> 
> for (i in 1:niter){
> 	
> 	m <- m + 1
> 	r <- 100*diff(log(wig20$CLOSE[1:m]))
> 	y <- r - mean(r)
> 	fit <- garch(y, order = c(1,1))
> 
> 	sink("garch21.txt", append = TRUE)
> 		summary(fit)
> 		logLik(fit)
>     	sink()
> 
> 	cv <- predict(fit, genuine=TRUE)
> 	fcv <- c(fcv,cv[length(cv)/2,1])
> 
> 	postscript("garch21.ps",encoding="ISOLatin2",
> 			title = paste("Day: ",wig20$DTYYYYMMDD[m]),
> 			paper = "a4",
> 			family = "URWTimes",
> 			append = TRUE)
> 		
> 		plot(fit, ask=FALSE)
> 		
> 	dev.off()
> 	
> }
> 
> sink("forecasts.txt")
> 
>      fcv
> 
> sink()
> 
> ###################################################

First, you should enclose the functions that output results in the loop in 
print(), so that they do print, in loops this has to be done and is a FAQ 
(often met with lattice graphics)

> 
> The most wondering thing for me is that, plots are stored in the ps 
> file, but only the ones from last iteration. In manuals there is:
> 
> append: logical; currently *disregarded*; just there for
>            compatibility reasons.
> 
> What does *disregarded* actually mean and when I will be able to add a 
> title before each set of plots, because now the option title is not 
> adding anything to file.
> 
Well disregarded here seems to mean exactly that. Why not open the 
postscript device before the loop starts, with onefile = TRUE, write to it 
in the loop, then dev.off() after the loop? You are getting the last 
graphic because that is what you are telling it to do, the others are 
being overwritten, because you are opening and closing the same file 
inside the loop, and append doesn't exist. 


> I will be very thankfull for any help.
> 
> Best regards,
> Wojtek Slusarski
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
> 

-- 
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Breiviksveien 40, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 93 93
e-mail: Roger.Bivand at nhh.no