issue-59 - add cross validation(cv) for tuning

[RaminNateghi]

Adding cross validation(cv) for tuning.

[cooperlab]

@RaminNateghi one thing that's missing here is to be able to group a set of experiments based on configuration. Example use case is to find the config with the best median over folds. Could be delayed for another PR.

[cooperlab]

Two suggestions:

1. For readability let's modularize the code to define operations on tables like sort by fold performance, global performance, or config performance. This approach will make it easier to extend and maintain. This allows us to return a table that is sorted in a sensible way and makes selecting the models trivial.
2. For sorting by config, we need a function to group table rows by config, then maybe add a column to the table enumerating the configs. From there it becomes easy to sort the table by config, and to calculate statistics for groups of rows.

[review-notebook-app[bot]]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB