#eda

Question 1: Single choice

The figure shows histograms of one feature, danceability, for the two classes of a song dataset (0 = song not liked, 1 = song liked). The two distributions overlap a lot, so on its own the feature separates the classes poorly. Should you remove it from the model?

Two overlapping histograms of danceability from 0.1 to 1.0. Class 0 (blue) peaks near 0.6, class 1 (orange) peaks near 0.7, and most of the two areas overlap.

No, because it could still be informative in combination with other features1
Yes, because a feature that cannot separate the classes alone carries no information2
Yes, because the range of its values is too narrow to be useful3
No, because a model simply ignores any feature that is not useful4

Timed test