R help - Oct 2007 - saving and loading complex objects

Dear Sir:

When I try to save large and very complex recursive objects with some components
containing models (such as the output from arima or lm ),
the resulting file sizes increase by 4 meg per save.

example directory:

amex                8 meg
argentina         12 meg
australia          16 meg
...

Moreover, I am unable to read these file objects back into R.
I note that readBin and writeBin are only for single mode data such as numeric.

Here is a section of my program or script:
  
SaveObj = function( x, nn ) {

pp=paste( "C://Program Files//R//R-2.5.1//arima//",nn,
sep="")

pp=paste( "save(  x, file=", sQuote(pp),")",
sep="" )
eval(parse(text=pp))
}	# end of function: SaveObj
load( file="C://Program Files//R//R-2.5.1//arima//amex" )




allArima=function( yy, popSize=10, generations=5 ){

nn = names(yy)
for( i in 1:dim(yy)[2] ) {

print( paste( "    iter: ", i, "name: ", " ",
names(yy)[i],sep="")  )

#						  data                 AVY              Note lagxy <-
CreateColumnSpace( yy, column = i, startLag = 1, endLag = 8, exclude=T, by=1 )

xx <- xy$xx

response = xy$response

t.index = xy$indx




# genetically select predictor column space

result = intGa( response =response, ... ,

                       FUN = LM, monitor = F, monitorFun = nn.Monitor, 

                       Evaluator = Std.Eval, wildCardPct = 0.10, 

                        parm1 = 4, parm2 = 1, parm3 = 3 ) 

x.model=result$bestModel

SaveObj( x=x.model, nn=nn[i] )

rm(result,x.model)

} 	# end for 


return( formulaLst )

}		# end of allArimaVery Sincerely,Lloyd Lubet

	[[alternative HTML version deleted]]

Why are you using eval and not:

path <- paste( "C://Program Files//R//R-2.5.1//arima//",nn,
sep="")
save(x, file = path)

?

It's possible that extra environments are getting saved.  (Someone
with more knowledge of eval might suggest a better solution, but that
might be a good place to start)

Hadley
>
> pp=paste( "save(  x, file=", sQuote(pp),")",
sep="" )
> eval(parse(text=pp))

On 10/1/07, Lloyd Lubet <lloyd.lubet at gmail.com>
wrote:
> Dear Sir:
>
> When I try to save large and very complex recursive objects with some
components containing models (such as the output from arima or lm ),
> the resulting file sizes increase by 4 meg per save.
>
> example directory:
>
> amex                8 meg
> argentina         12 meg
> australia          16 meg
> ...
>
> Moreover, I am unable to read these file objects back into R.
> I note that readBin and writeBin are only for single mode data such as
numeric.
>
> Here is a section of my program or script:
>
> SaveObj = function( x, nn ) {
>
> pp=paste( "C://Program Files//R//R-2.5.1//arima//",nn,
sep="")
>
> pp=paste( "save(  x, file=", sQuote(pp),")",
sep="" )
> eval(parse(text=pp))
> }       # end of function: SaveObj
> load( file="C://Program Files//R//R-2.5.1//arima//amex" )
>
>
>
>
> allArima=function( yy, popSize=10, generations=5 ){
>
> nn = names(yy)
> for( i in 1:dim(yy)[2] ) {
>
> print( paste( "    iter: ", i, "name: ", " ",
names(yy)[i],sep="")  )
>
> #                                                 data                 AVY 
Note lagxy <- CreateColumnSpace( yy, column = i, startLag = 1, endLag = 8,
exclude=T, by=1 )
>
> xx <- xy$xx
>
> response = xy$response
>
> t.index = xy$indx
>
>
>
>
> # genetically select predictor column space
>
> result = intGa( response =response, ... ,
>
>                        FUN = LM, monitor = F, monitorFun = nn.Monitor,
>
>                        Evaluator = Std.Eval, wildCardPct = 0.10,
>
>                         parm1 = 4, parm2 = 1, parm3 = 3 )
>
> x.model=result$bestModel
>
> SaveObj( x=x.model, nn=nn[i] )
>
> rm(result,x.model)
>
> }       # end for
>
>
> return( formulaLst )
>
> }               # end of allArimaVery Sincerely,Lloyd Lubet
>
>         [[alternative HTML version deleted]]
>
> ______________________________________________
> 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.
>

-- 
http://had.co.nz/