lets lay out what this is:
so this is a app that will let benchmark a few different active ML approaches. it will also allow a user to play against various active ML algorithms. i.e. can i beat it?
what this means in terms of code: so i will need a dataset, preferably a smallish one so its fast and not overwhelming. test train split. Test: This is used to assess the predictive performance of models trained on currently labelled data train: this provides some initial seed data, and is the space that users and AML can select from to label.
The process will look like a set of batches, say multiple rounds of chosing 5 things. After each batch, a model is trained and metrics plotted.