style: apply ruff formatting and fix label mapping for #589

[mariam851]

Hi @rasbt,

I hope you are doing well.

I have submitted this PR to address issue #589, where plot_decision_regions was missing colors when dealing with non-consecutive or negative class labels (e.g., [-1, 1]).

The Problem: The function previously relied on the actual class label values as indices for color mapping, which caused issues when labels were not starting from 0 or contained negative integers.

The Solution: I modified the internal logic to map arbitrary class labels to a zero-indexed range (0, 1, 2...) specifically for the coloring and contouring process. This ensures that:

Decision regions are always colored correctly regardless of the label values.

The original labels are preserved for the scatter plot and legend to maintain data integrity.

Changes made:

Introduced a label_to_index mapping.

Mapped Z predictions to Z_indexed for the contourf and contour calls.

Verified the fix with a Perceptron classifier using [-1, 1] labels.

Applied Ruff for formatting and linting to ensure code quality.

Fixes #589

[rasbt]

Thanks for the PR(s)! And sorry for the late reviews and responses. Mlxtend is an older, lower-priority project, and I am currently a bit swamped with other things, so I can't make as much time for it that I would like to. Thanks for trying to improve it via the PRs, though!

[rasbt]

Tested it locally, and it works fine. Thanks!

[mariam851]

Thank you for the review and merge. I appreciate your feedback and time, and I’m glad the fix was helpful.