As a statics researcher, I've been using R since university and I know it quite well, I also know that it's immediate, but it quickly gets chaotic, and this also happens because of the variety and inconsistency of the packages and object types.
I've came to know scikit-learn some years ago, and since then I've always used that package for machine learning, it's so neat and really helps you doing things in the right way! I learned to use pandas, which is very nice but too unpredictable and also quite bugged, and some matplotlib, which is indisputably versatile, but frankly, so hard and time consuming!
For these and further reasons still, I would like to move back to R, but I know I would miss scikit-learn. I never really liked caret, but I have to admit I never really gave it a chance, mlr3 seems more promising, but maybe mlr is a good option too.
One good thing I want to have is fully customizable pipelines, that can take in custom functions or classes. This is fairly easy in python.
Do you have any recommendations about where to look, what to learn? pros and cons over python tools?
