[R] Missing Data in Stepwise selection of Logistic regression

[Michael Dewey]

At 07:02 24/02/2012, Kawthar Alajmi wrote:
>Hi all,
>
>I am running Stepwise logistic regression and i have :
>1- Multiple covatiates included in each model (No missing data)

So there is no missing data on any covariate?

>2- Genotype data (SNPs) about 500,000 .
>I partitioned the data to multiple files (there are missing data)

And now there is missing data on some of the covariates?

I suggest you revisit how you partitioned your dataset.

I also suggest you get some local expert advice on the risks involved 
on doing half a million logistic regressions, let alone half a 
million stepwise regressions.


>I run the step by including all the covariates and one SNP at each model.
>
>but i got this message :
>
>  number of rows in use has changed: remove missing values?
>In addition: There were 37 warnings (use warnings() to see them)
>
>How to overcome this problem ??
>
>         [[alternative HTML version deleted]]

Michael Dewey
info at aghmed.fsnet.co.uk
http://www.aghmed.fsnet.co.uk/home.html