Observable Framework View source

Penguin classification

Analysis

The logistic regression failed to classify individuals. Let’s check what was amiss, with this faceted chart:

As we can see in the top-right facet, the classifier could have done a better job at discovering that the Torgensen island only hosts penguins of the Adelie species.

We could try re-running the analysis with different options — e.g., LogisticRegression(solver = "newton-cg") — to see if that results in better predictions. (Spoiler: it does!) See the scikit-learn documentation for details.