Exercise

Coarse to Fine Iterations

You will now visualize the first random search undertaken, construct a tighter grid and check the results. You will have available:

  • results_df - a DataFrame that has the hyperparameter combination and the resulting accuracy of all 500 trials. Only the hyperparameters that had the strongest visualizations from the previous exercise are included (max_depth and learn_rate)
  • visualize_first() - This function takes no arguments but will visualize each of your hyperparameters against accuracy for your first random search.

If you wish to view the visualize_first() (or the visualize_second()) function definition, you can run this code:

import inspect
print(inspect.getsource(visualize_first))

Instructions 1/3

undefined XP
    1
    2
    3
  • Use the visualize_first() function to check the values of max_depth and learn_rate that tend to perform better. A convenient red line will be added to make this explicit.