parallel back-end

[moezali1]

trafficstars

At the moment scikit-learn uses loky backend to single node parallel processing. There are 3-4 other options for backend like multithreading or ray. We should create a new parameter in setup for the user to set that backend.

We can set the backend that gets reflected in the n_jobs parameter by using a function from sklearn.utils module.

Let's call this new parameter in setup: parallel_backend

[revanks]

HI @moezali1 When can we expect Pycaret with parallel back-end. It would be great if it is available asap to use it in production codes, and Thanks lot for this awesome library.

[jmakov]

Is there any example how to run ray to compare models or train on a cluster?

[jmakov]

Is there any update on this or is ray not planned to be supported?