Yellowbrick is a Python visualization library for machine learning. It extends the Scikit-Learn API to provide visual diagnostic tools for classifiers, regressors, clusterers, transformers, pipelines, feature extraction tools and more. This tag should be used to ask questions about how to use visualizers, how to extend or modify visualizations, or how to interpret diagnostics. This tag is commonly used with the scikit-learn and matplotlib tags.
Yellowbrick (sometimes referred to as scikit-yellowbrick) is a Python library that extends the Scikit-Learn API to enhance the machine learning workflow with visual diagnostics with matplotlib. The yellowbrick tag is therefore usually applied in combination with the scikit-learn, python, and matplotlib tags. Good questions for this tag include:
- Questions about how to work with or extend existing visualizers
- Questions about how to interpret visual results
- Questions about how to modify resulting figures or annotate them
- Questions about how to create new visualizers
The best questions will include a code example along with the figure being generated by Yellowbrick. In order to allow others to run your code, if the visualization is not data specific (e.g. the questions is not the result of specific input), please use one of the example datasets from the Yellowbrick tutorial or one of Scikit-Learn's dataset generation methods. It is also very helpful if you include the version of Yellowbrick you're using, which can be found with print(yellowbrick.__version__).
For result interpretation questions, please be as general as possible and focused on the content of the visualizer. For example, a good question is "what is the meaning of the macro-average curve on ROC/AUC with more than two classes?" A poor question is "how do I make my model have a higher F1 score?"
Finally, for both folks asking questions and those responding, Yellowbrick contributors and developers take seriously respectful discourse. In addition to be nice, the StackExchange code of conduct, Yellowbrick also follows the Python Software Foundation Code of Conduct
Definitions
Yellowbrick extends the Scikit-Learn API with a new Estimator called a Visualizer. Visualizers are estimators, in that they can be fit with data in order to produce a visualization. Anything that produces a visualization in Yellowbrick is a Visualizer, though there are different types.
A FeatureVisualizer produces a representation of the feature space or data space. These are used to explore the input to models or the relationship of data to the model.
A ModelVisualizer produces a representation of the model space, describing how the model interacts with data or behaves. It does this in two ways, by describing internal parameters of the model, or by describing the relationship to test data with a ScoreVisualizer.
