From my time developing Health AI, I have found some things that we could have accomplished better with data tracking and model performance tracking. This repo will help target this in a way that allows it to be integrated with systems in a more pythonic way.
This system needs to allow the tracking of model progress as well as the date and time of training. This also will mean that I want to be able to perform some analysis on different metrics and time sequence tracking data.
I also want systems that look at the datasets that are being used and can analyze for any overlapping of data in the training and testing set based off the name of the file or some instance information to avoid data leakage.
This system also needs to hold certain files and organize them to correlate research decisions and for ML work in some way.