Following a seminar presentation, a member of the audience approached me to follow up on a point I had addressed. He explained that he worked in a bank, and also taught a graduate-level data mining course. He asked if I had any advice about how to turn the business description of a problem into a form that could be answered with data mining and data. Indeed, several other people in the same group asked the same question. Since then, I have been asked what is essentially the same question many times and in many places ranging from email discussions to audience questions at academic conferences to business tutorial presentations to individuals engaged in trying to discover how to mine their data. It has become obvious that this is a pressing topic: How can real world business problems be formulated so that data mining can answer them? But beyond that is an even more fundamental question: What problems can data mining usefully address—and how?
Discovering those areas to which data mining can be applied most effectively, and then translating business problems into a form that can be addressed by data mining and modeling, is very similar to a problem that faces almost everyone studying algebra for the first time—the dreaded word problems. How do you turn a statement in words into what is essentially a mathematical formula? In a sense, this book is a long answer to those questions—how to use, where to fit, and how to get the most value out of mining and modeling in strategic and tactical business applications.