Improving the plot
In order to make the plot more readable, we need to do achieve two goals:
- Re-order the bars in ascending order.
- Add labels to the plot that correspond to the feature names.
To do this, we'll take advantage of NumPy indexing. The .argsort() method sorts an array and returns the indices. We'll use these indices to achieve both goals.
This exercise is part of the course
Marketing Analytics: Predicting Customer Churn in Python
Exercise instructions
- Calculate the sorted indices of
importancesby usingnp.argsort()onimportances. - Create the sorted labels by accessing the columns of
Xand indexing bysorted_index. - Create the plot by indexing
importancesusingsorted_indexand specifying the keyword argumenttick_label=labels.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Sort importances
sorted_index = ____(____)
# Create labels
labels = X.columns[____]
# Clear current plot
plt.clf()
# Create plot
plt.barh(range(X.shape[1]), importances[____], tick_label=____)
plt.show()