better hyperparameter optimization using evolutionary algorithms through sklearn-deap

[ClimbsRocks]

one optimization we can make is to run grid search when the search space is small (such as when we're comparing model types), and run evolutionary algorithm search (EAsearch) when the search space is large (deep learning architecture!)

[ClimbsRocks]

i think this is largely implemented.

now we need to figure out a few things:

1. if we want to report more analytics on the results of GS/EASCV
2. creating more in-depth param distributions to search over for some of the popular algorithms
3. uhhh, i think that's it? maybe figure out if we want to adjust verbosity?
4. definitely a larger search space for deep learning (lots more params)
5. there was at least one more thing

[ClimbsRocks]

we'll probably want to tweak our search settings. i think we might search too small of a space initially.