4 Comments

Very insightful , thanks for sharing

Expand full comment

Bayesian stats is taking machine learning

to another level in most aspects. Excellent article.

Expand full comment

Thanks for sharing this. Yet another concept least talked about.

Expand full comment

Insightful article, Avi Chawla! I have observed similar results. In the past, I have used scikit-optimize BayesSearchCV and was really pleased with the results, but unfortunately it has not been maintained or updated since September 2020 and the project seems to be abandoned. What library do you recommend for implementing Bayesian Optimization in Python?

Last year, I created a discussion on GitHub proposing an implementation of BayesSearchCV in scikit-learn. Unfortunately, that does not seem to be the direction that the scikit-learn team wants to go in. One recommendation in the discussion was to use HalvingSearchCV. Curious to know your thoughts on that - performance, speed, etc.

GitHub scikit-learn discussions:

https://github.com/scikit-learn/scikit-learn/discussions/26141

https://github.com/scikit-learn/scikit-learn/issues/26170

Expand full comment