Pushed the Other Way

A scaled logistic regression reads 30 tumour measurements and predicts class 1 (benign) or 0 (malignant). A positive push moves a row towards benign; a negative push moves it towards malignant.

For one test sample, a doctor wants both sides of the story.

Task: write both_sides(test_index, n).

• Load load_breast_cancer(return_X_y=True, as_frame=True).
• Split with train_test_split(X, y, test_size=0.25, random_state=0, stratify=y).
• Fit make_pipeline(StandardScaler(), LogisticRegression(max_iter=1000)) on the training rows.
• Take test row X_test.iloc[[test_index]]. Its pushes are its scaled values times coef_[0].
• Return a dict:
  • "predicted": the predicted class, a plain int
  • "benign": the names of the n columns with the largest positive pushes, largest first
  • "malignant": the names of the n columns with the most negative pushes, most negative first