The Loud Column

A fruit stall describes each fruit by [weight in grams, sweetness from 1 to 10] and labels it "sour" or "sweet". Weight is written in much bigger numbers than sweetness, so without scaling it drowns sweetness out.

Write bare_and_scaled(rows, labels, new_row, k).

• Bare: fit KNeighborsClassifier(n_neighbors=k) on the rows as they are.
• Scaled: fit make_pipeline(StandardScaler(), KNeighborsClassifier(n_neighbors=k)) on the same rows.
• Ask both for the label of the single fruit new_row.

Return a dict {"bare": ..., "scaled": ...} holding the two predicted labels as strings.