Predicting outcome with multiple groups

Hi everyone,

I'm currently working with 3 different groups (3 disease populations) and I would like to predict which of the three groups they fall into based on a measurement we did. I was thinking that it should be something like multinominal logistic regression but how can I then get a vector out to predict the response in the future?


Thanks a lot for your comment hlsmith!

I would like to use the dataset I have now to predict new cases in the future. For example, predicting which one of the three groups a new subject would be classified as if I recruited one more subject and wanted to know if that one had disease 1, 2 or 3.

Thanks in advance,



Less is more. Stay pure. Stay poor.
So what is the "measurement" you did. So do you have a continuous variable you are looking to use to trichotomize patients with? Or do you have multiple predictors?
Yes, I'm planning to use electrophysiological measures (continuous variables). It would be interesting to use multiple predictors but first of all it's about one predictor.

Thanks a lot,