• The custom notification program can serve 4000 people. The program cost is $25 per person. The social cost of a False Positive, is $50 per False Positive. The social benefit of a True Positive will be either $4,000 or $400 depending on circumstances.

• The summer jobs program can serve 700 people. The program cost is $300 per person. The social cost of a False Positive is $2 per False Positive. The social benefit of True Positive will be either $3,000 or $300 depending on circumstances

We ideally would like to optimize models so that they incorporate some aspect of the problem we are trying to address. For example, the custom notification program above can target 4,000 people and the public health program can serve 700 people. Describe how one could incorporate some of this information into the model selection process?