I've a pooled group of individuals and a given number of features. So that my matrix looks like:

individuals feature1 feature2 feature3
bob 1 0 1
ralph 0 1 1
mark 1 0 1

I want to discriminate between bob, ralph and mark. I'm not interested much in the features, I just want to know this line is ralph and not bob or mark. And I want to be quite accurate about it.

Could I use Naive classifier for this? I'd like to use because I may have some prior knowledge to include. Like 50% is bob, 25% is ralph and 25% is mark.

In this case I won't have a training and test data. Could I train the classifier on this data and use it to deconvolute as well? I won't use it in prediction on other data, but just to deconvolute this dataset. If not, can I get some reading to understand why?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.