The Clinic Report

A doctor does not want an R² alone. They want to know how far out the model usually is, and whether that is any better than simply guessing the average.

Write clinic_report(columns).

• Load the data with load_diabetes(return_X_y=True, as_frame=True) and split it with train_test_split(X, y, test_size=0.25, random_state=0).
• Keep only the columns named in columns.
• Fit make_pipeline(StandardScaler(), LinearRegression()) on the training rows and predict the test rows.
• Also make a lazy baseline: predict the mean of the training answers for every test row.

Return a dict, with every number a plain float:

• "r2": the model's test R², rounded to 3 places.
• "rmse": the model's test RMSE (square root of the mean squared error), rounded to 1 place.
• "mae": the model's test mean absolute error, rounded to 1 place.
• "baseline_rmse": the lazy baseline's test RMSE, rounded to 1 place.