8000 Support clusterers in `DecisionBoundaryDisplay` · Issue #33119 · scikit-learn/scikit-learn · GitHub
[go: up one dir, main page]

Skip to content

Support clusterers in DecisionBoundaryDisplay #33119

@ThexXTURBOXx

Description

@ThexXTURBOXx

Describe the workflow you want to enable

So far, response_method="auto" does not resolve correctly (list of options as for classifiers instead of predict).
The approach I implemented is: using PCA, followed by DBSCAN and Logistic Regression as a final step.
Properly showing the relevant datapoints, categorised/colored by their identified clusters, would be great without an extra call to scatter.

Describe your proposed solution

As per discussion linked below, proper support for clusterers should be added to DecisionBoundaryDisplay.

Describe alternatives you've considered, if relevant

Simply plotting a scatterplot on top does not necessarily yield the same class colors, because the clusters are not necessarily ordered in the same way as the y values.

Additional context

#33015 (comment)

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions

      0