Exercise

See how the ensemble performed

Let's check performance of our ensembled model to see how it's doing. We should see roughly an average of the R\(^2\) scores, as well as a scatter plot that is a mix of our previous models' predictions. The bow-tie shape from the custom loss function model should still be a bit visible, but the edges near x=0 should be softer.

Instructions

100 XP
  • Evaluate the R\(^2\) scores on the train and test sets. Use the sklearn r2_score() function (already imported for you) with train_targets and train_preds from the previous exercise.
  • Plot the train and test predictions versus the actual values with plt.scatter().