Taylor Miller

When instantiating trainer guess which label is positive class for binary classifications

1

Using the [`model_eval.gues_binary_positive_label()`](url) method. Randomzied searches using `roc_auc` sometimes fail since there is no positive class label specified. I'm uncertain how to roll this down through the randomized search call......

## Background - With the multiclass MVP all metrics were changed to default to accuracy. This isn't ideal. - The trainer should detect what kind of classification task and use...

# Benefits We outsource categorical encoding. ## Current methods in the library: - Ordinal - One-Hot - Binary - Helmert Contrast - Sum Contrast - Polynomial Contrast - Backward Difference...

Code coverage service: CI integration

3

@mxlei01 Would you take this on once #344 is complete?

# short term urgent fix - refactor out the list that is passed to the plotter - have the plotter limit the list length so that the full list can...

This requires #458 to be fixed first - investigate if it's possible to keep the bars a constant height and scale the plot as the `feature_importance_limit` increases. - For example,...

## Background This is hard to read - Switch this plot to a horizontal bar plot

While this would benefit from #446 it could be done with a bit of hacky logic as an MVP ## GWT - Given I am a user that has a...

Handle a missing target or grain column more elegantly when instantiating trainer

7

## TODO - [ ] grain column should not be required. - [ ] handle missing target column with a nice error - [ ] tests for both situations ```...