Training progress verbosity

[gjdv]

It would be nice to have a bit more insight into the training progress (ETA etc). For a grid search, one can set a verbosity level that is taken into account by sklearn, but sklvq does not provide such a verbose argument.
To serve my needs, I have solved it with a callback class like:

from tqdm import tqdm
class ProgressTracker:
    def __init__(self, max_runs):
        self.run_tracker = iter(tqdm(range(max_runs + 1), desc="Training progress"))

    def __call__(self, state):
        next(self.run_tracker)
        return False

This prints e.g.,:

Training progress:  11%|█         | 1115/10001 [00:45<06:01, 24.57it/s]

Perhaps you can consider adding some functionality like this to the LVQBaseClass class such that calling fit(verbose=1) prints this (not sure if it is possible to have multiple callback objects in the solver_params or whether there is a central place that is visited for each run where this code could reside); or alternatively, add the above as an example in the example usages / documentation.

[rickvanveen]

Adding it to the docs seems the best solution (for now). I will have a look after updating some other things.