[R] Re gression between columns in matrix - with some/full NAs

[rcoder]

Hi Anna,

Thanks for your reply and suggestions. I'm trying something different, based
on regression this time, albeit giving similar problems. I want to regress a
single column in a matrix with each other column in the same matrix, and o/p
the intercept and slope coefficients to a results matrix. 

The loop below works, even when I have columns with partial NAs, but exits
with an error when a column with only NAs is encountered. So, the
'na.action=NULL' statement (see last line of code below) doesn't seem to
work in this case, or perhaps I am applying it incorrectly. I would be very
grateful for any pointers in the right direction:

>tt<-time(SourceMat)
>ResultMat<-matrix(NA, ncol=colnum, nrow=rownum)     #creates an o/p
template matrix

#loop through each column in the source matrix:
>for (i in 1:5000)
		{
	sel_col<-[col(SourceMat)==i]	#selecting the correct column in the matrix in
turn
	SourceMat[,i]<-coef(lm(tt~sel_col), na.action=NULL)
		}


Thanks,

rcoder



rcoder wrote:
> 
> Hi everyone,
> 
> I'm having trouble applying the Cor() function to two matrices, both of
> which contain NAs. I am doing the following:
> 
> a<-cor(m1, m2, use="complete.obs")
> 
> ... and I get the following error message:
> 
> Error in cor(m1, m2, use = "complete.obs") : 
>   no complete element pairs
> 
> Does anyone know how I can apply a correlation, ignoring any NAs?
> 
> Thanks,
> 
> rcoder
> 

-- 
View this message in context: http://www.nabble.com/correlation-between-matrices---both-with-some-NAs-tp18721853p18739148.html
Sent from the R help mailing list archive at Nabble.com.