Briefly, Haimowitz and Schwarz took this new knowledge and cycled back to the beginning of the data mining process. They used the knowledge to define a precise predictive modeling problem: using data that are available at the time of credit approval, predict the probability that a customer will fall into each of these clusters.
