R help - Nov 2010 - Problem with 'lars' package

Hello,

I have problems with 'lars' package. I found the previous post of the
person
who had the same issue, but the suggested solution in that post did not help
me.

I created the matrices:

ResponseMatrix <- data.frame (GAOdecision=GAOdecision)
PredictorsMatrix <- data.frame (WeaponvsNon = WeaponvsNon, ProductvsService
= ProductvsService, KDuration = KDuration, BusinessSize = BusinessSize,
Bidders = Bidders, Staging = Staging, Criteria = Criteria, KPricing KPricing,
Political = Political)

and then
output <- lars(PredictorsMatrix, ResponseMatrix)
*Error in one %*% x : requires numeric/complex matrix/vector arguments*
*
*
I did not create any variables x in my code. I also used rm(x) in the
beginning of my code, but it did not help.

I would really appreciate any help.

Thanks,
Vlad

	[[alternative HTML version deleted]]

Hi,

On Fri, Nov 5, 2010 at 4:14 PM, Vladimir Subbotin
<subbotinvladimir1 at gmail.com> wrote:
> Hello,
>
> I have problems with 'lars' package. I found the previous post of
the person
> who had the same issue, but the suggested solution in that post did not
help
> me.
>
> I created the matrices:
>
> ResponseMatrix <- data.frame (GAOdecision=GAOdecision)
> PredictorsMatrix <- data.frame (WeaponvsNon = WeaponvsNon,
ProductvsService
> = ProductvsService, KDuration = KDuration, BusinessSize = BusinessSize,
> Bidders = Bidders, Staging = Staging, Criteria = Criteria, KPricing >
KPricing, Political = Political)
>
> and then
> output <- lars(PredictorsMatrix, ResponseMatrix)
> *Error in one %*% x : requires numeric/complex matrix/vector arguments*
This is likely talking about variables in the function -- for
instance, not that the first parameter in the function is named 'x'.

Also, your ResponeMatrix and PredictorsMatrix should be of type
*matrix* not data.frame ... you can try to convert it with
as.matrix(ResponseMatrix). You should then ensure that all values in
your response and predictors matrix are numeric, ie:
all(is.numeric(ResponseMatrix)), etc ... also check for odd values
like NA, NaN or Inf.

Lastly, consider using the glmnet package instead of lars. It provides
a superset of the functionality from lars, and its "core" is written
in fortran, so will likely be much faster and deal with larger
problems. It's also more actively maintained (note a new version of
glmnet was just released this week).

-steve

-- 
Steve Lianoglou
Graduate Student: Computational Systems Biology
?| Memorial Sloan-Kettering Cancer Center
?| Weill Medical College of Cornell University
Contact Info: http://cbio.mskcc.org/~lianos/contact