R help - Jun 2009 - grep on vectors?

Input: dataframe with 300+columns for a regression. It consists of sets of
factors whose names have the same structure. For example, aa1,aa2,aa3 could be
one set of factors.

After reading in the dataframe, I would like to compute the density (%nonzeroes)
for certain groups of factors and delete the factors which are below the density
threshold. I would like to use regular expressions to specify the factor names.

density.factor = c("^aaa","^bbb")
density.faccol=c()
for(fac in density.factor){
    density.faccol=c(density.faccol,grep(fac,names(data.df)))
}
data.df=data.df[,-density.faccol]

Is there a way to avoid the for loop? The following seems to work:
  lapply(density.factor,grep,names(data.df))
However, that produces a list of lists which need to be merged. Note that in the
above example since we have 2 regular expressions, there will be two lists but
in the general case there will be many more.

Questions (i) how do I merge the lists into a single list (ii) is there a better
way to achieve the "vectorized" grep?

Thanks.

On 30/06/09 17:53, Chuck White wrote:
> [...]
>
> Is there a way to avoid the for loop? The following seems to work:
>    lapply(density.factor,grep,names(data.df))
> However, that produces a list of lists which need to be merged. Note that
in the above example since we have 2 regular expressions, there will be two
lists but in the general case there will be many more.
It is hiding, not avoiding the for loop, but if you are happy with the 
lapply() approach then just use unlist() on the result:

unlist(lapply(density.factor, grep, names(data.df)))

I wouldn?t worry about optimizing performance: it isn?t the sort of 
thing you are going to be running a million times per second.  Keep it 
understandable and maintainable.

Hope this helps a little.

Allan.