Optimization driven data mining and credit scoring

Assume we are given a large collection of objects, each with several hundred attributes, and we wish to assign scores and take appropriate actions for each object in such a way as to maximize a given objective function defined on the entire collection. In this paper, we describe a methodology that uses data mining to divide the objects into different clusters on the basis of their attributes so that individualized scores and appropriate actions can be assigned to the objects in each cluster. This problem arises in a variety of applications: For example, we may wish to assign a credit score to a credit card prospect indicating the likelihood that the prospect will make credit card payments and then to set a credit limit for each prospect in such a way as to maximize the over-all expected revenue from the entire collection of prospects. In the terminology we will use in this paper, the credit score is called a statistical attribute and the credit limit a control attribute. The methodology we describe below uses data mining to better estimate the statistical attributes and to better set the control attributes. As another example, consider the real-time detection of fraudulent credit card transactions. In this example, the goal is to assign a risk score