Two models predicted the same delivery times. Model A's predictions are guess_a and model B's are guess_b. The true times are truth. Which one is better depends on how you measure.
Write better_model(truth, guess_a, guess_b, metric). metric is one of:
"mae": lower mean absolute error wins."rmse": lower root mean squared error wins."r2": higher R² wins.Return "A" or "B". Use the sklearn.metrics functions. The two models never tie on the metric asked for in the tests.