Question

I am trying to run a stepwise regression in R with 600 odd variables contained as the column name from the header of a .csv file.

How do i use the column name as variables in a regression equation?

I am very new to this and my limited understanding it that i can save the column as a list and use it for running a glm eg

model.1 <- glm(x~ paste(list), family= poisson, link = logit).

Any help is highly appreciated. thanks in advance

Answer 1

If you have read your data in correctly (e.g. with header=TRUE as specified in the comments above), you should end up with a 600+-column data frame (1 column for the x response, and a column for each predictor variable): I will call this mydata for now. In that case as @TylerRinker suggests you could just include all the predictors: glm(x~.,data=mydata,family=poisson) (the logit link is the default link; if you want to specify it explicitly you can say glm(x~.,data=mydata,family=poisson(link="logit")). You could then use step, or stepAIC from the MASS package.

However, I have to add that unless you know what you’re doing, stepwise regression on 600 variables is a really, really, really BAD idea from a statistical point of view (Google something like “stepwise regression problems” or “stepwise regression Harrell”). I would strongly encourage you to take a look at something like the glmnet package, which takes a more sensible approach to modeling with lots of predictors.