|
Abstract:
|
Most managers agree that data mining and modeling play an important role in assuring success of a marketing program. However, errors can creep into the data mining process---problems that both novice managers and experienced analysts can confront. And these errors are more often nontechnical in nature. This paper will cover the author's many years of experience as both a marketer and statistician as they relate to model building and the associated nontechnical problems that typically arise, specifically in the database marketing arena. We will discuss what these error-prone situations are and simple solutions to confront them. Areas prone to error include defining objectives correctly, using appropriate predictors, quality control procedures, and many other seemingly mundane activities. The problems discussed will be framed through actual marketing problems.
|
