I want to predict a vector whose elements are binary. What is the most promising machine learning model for doing this at present? Is there a corresponding Julia package?
That’s a pretty broad question - see List of Supported Models · MLJ and read a bit about the classifier models to see if any might be particularly suited, or if you don’t want to spend any time just use Xgboost Classifier and a logistic regression via GLM.jl
4 Likes
Thank you, @nilshg !
What is the relationship between Julia’s different machine learning packages? (For example, MLJ.jl, Flux.jl, …) Is there a kind of overall introduction like an overview chart?
Not as far as I’m aware. MLJ is a framework which unifies access to lots of different models and provides convenience functions for tuning, stacking etc., so in that sense it sits on top of a number of ML packages which implements a specific methodology/estimator.
1 Like
Thank you very much, @nilshg ! Now I have general cognition of where MLJ.jl is located.
If you want to explore probabilistic methods as well, I recommend RxInfer.jl and Turing.jl. You could start with simple categorical models and build out more complex models from there.
1 Like
Thank you @langestefan I’ll have a try if possible. 