Hello everyone!
I am trying to figure out how to do a classification task using the pymc3 library. I have read an amazing article on how to achieve that but the article caters to the classification problem having only two features (target variable included). In the article, the author samples one value for beta and one value for alpha (as he has one independent variable). However, in my case, I have a few too many variables. How should I go about sampling betas and alphas? Should I sample betas and alphas for each of my features? If yes then what is the proper way to do that? Any kind of help would be great!
Cheers!
Conventionally, you would have a single intercept (alpha) and one slope (beta) per predictor/feature. But there are many, many different ways to construct such model. This notebook covers logistic regression and this one covers binomial regression. I might suggest starting with those.
3 Likes
Fixed!