MATLAB: Using the Classification Learner to train a Multinomial Logistic Regression

Question

I am interested to use the Classification Learner app to train a multinomial logistic regression model, specifically, for ordinal response.

I was only able to find logistic regression option (binomial) within Classification Learner.

I have also used mnrfit for this purpose, but was wondering if there is a difference in performance.

Thanks

Answer 1

Hi,

Currently Classification Learner App has Decision Trees for multi-class classification problems. As you mentioned mnrfit does the Multinomial Logistic Regression, you may use glmfit with corresponding distribution for doing the same. You can find an example here. For more Information you may take a look to the following documentation.

Multinomial Models for Nominal Response