I am rebuilding the "framework" I wrote for doing my MSc dissertation on the classification of fake reviews. Now that I have the time, the code is far better or

I am rebuilding the "framework" I wrote for doing my MSc dissertation on the classification of fake reviews. Now that I have the time, the code is far better organized and the higher level structure pretty clean in that I can copy a module, customize it a bit and now a different feature set can be generated. Churning out different ensembles should be fairly easy as well. And of course I'm throwing dask, mlflow and dvc into the mix... I don't really need to do this (ought to be looking for a job...) <and I'd rather write code than pay to use someone else's SaaS> but now that I have quite a number of different review datasets to play with, a framework that lets me spin up 10+ processes across the 5 machines I have at my disposal will make short work of most the sets. I will need the uni's HPC cluster to deal with the Amazon dataset, though. Using dask will surely help with that. Then it's writing yet another paper... Not being allowed to go outside is rather motivating.