idanmoradarthas / DataScienceUtils / 18333157122

100%

push

github

Replace box plot with violin plot for numeric vs categorical interaction in plot features interaction (#69)

* Change visualize feature for float from distplot to violinplot

* CR fix

* Minor formatting

* Force install tk in windows

* revet

* trying to fix tk in windows

* Set Backend for Matplotlib via Environment Variable in GitHub Actions

* Refactor plotting functions in preprocess.py for improved clarity and functionality

- Updated plotting functions to streamline feature comparisons, including renaming and consolidating functions for categorical and datetime features.
- Removed unnecessary DataFrame duplication in plotting functions.
- Enhanced the handling of datetime features in plots for better visualization.

* reformatt using ruff

* Update dependencies

* Enhance feature interaction plotting in preprocess.py

- Added parameters to control outlier inclusion and IQR multiplier for violin plots in categorical-vs-numeric scenarios.
- Updated documentation to clarify behavior based on feature types and added type-aware defaults for plotting functions.
- Improved handling of outliers in categorical-vs-numeric plots for better visualization.

* Update failed images

* Update images

* Refactor code formatting in visualize_feature function for improved readability

* Revert change

* Update documentation for feature visualization

- Changed the description of the plot type from boxplot to violin plot for numeric vs. categorical feature relationships, emphasizing the combination of box plot and kernel density estimate.
- Removed outdated code example for looping through features, streamlining the documentation.
- Deleted an unused image related to feature visualization.

* Update documents

* Update requirments

* Update again

* Add test for interaction plot between numeric and categorical features without outliers

- Introduced a new test to validate the interaction plot for two numeric features while excluding outliers.
- Ensured the plot s... (continued)

66 of 66 new or added lines in 1 file covered. (100.0%)

584 of 584 relevant lines covered (100.0%)

12.0 hits per line