The raw materials of the financial analysis are completed. However, in this activity, your aim is to generate some additional insights from these results, to provide the client with more context around how the predictive model we built can generate value for them. In particular, we have looked at results for the testing set we reserved from model building. The client may have more accounts than those they supplied to us, that are representative of their business. You should report to them results that could be easily scaled to however big their business is, in terms of the number of accounts.

We can also help them understand how much this program will cost; while the net savings are an important number to consider, the client will have to fund the counseling program before any of these savings will be realized. Finally, we will link the financial analysis back to standard machine learning model performance metrics.

Once you complete the activity, you should be able to communicate the initial cost of the counseling program to the client, as well as obtain plots of precision and recall such as this:

Expected precision-recall curve

This curve will be useful in interpreting the value created by the model at different thresholds.

Note: Additional steps to prepare data for this challenge, based on results obtained in a previous lesson, have been added in the Notebook.