Fix unreadable labels in plot_confusion_matrix in case of imbalanced data and show_normed=True

[dynobo]

Change

If show_normed is True, use the normed conf mat for background coloring instead of the absolute conf matrix, because label coloring (black/white) also follows normed conf matrix. This fixes unreadable labels that can occur with imbalanced datasets.

Description

If arg show_absolute=True and show_normed=True the background-color is choosen by the absolute value, while the label color is chosen by the normed value. This can lead to unreadable labels for imbalanced dataset.

Before / After:

Related issues or pull requests

n/a

Pull Request Checklist

• Added a note about the modification or contribution to the ./docs/sources/CHANGELOG.md file (if applicable)
• Added appropriate unit test functions in the ./mlxtend/*/tests directories (if applicable)
• Modify documentation in the corresponding Jupyter Notebook under mlxtend/docs/sources/ (if applicable)
• Ran nosetests ./mlxtend -sv and make sure that all unit tests pass (for small modifications, it might be sufficient to only run the specific test file, e.g., nosetests ./mlxtend/classifier/tests/test_stacking_cv_classifier.py -sv)
• Checked for style issues by running flake8 ./mlxtend

[coveralls]

Coverage remained the same at 91.596% when pulling 2a36c91 on dynobo:master into cce9e60 on rasbt:master.

[rasbt]

Great catch, thanks a lot!

[dynobo]

Typo fixed and also added link to PR

[rasbt]

LGTM! Thanks a lot!