One widely publicized competition in recent years has been the Knowledge Discovery in Databases (KDD) Cup competition. In 2009, KDD distributed data for the classic marketing problem of churn (i.e., predicting which customers will end their relationship with a business and choose a competitor). Recently, a team of modelers from SAS decided to use SAS Enterprise Miner software and the 2009 KDD data to build a highly accurate churn model. The software has extensive capabilities for a comprehensive data mining or machine learning process.
As with much real-world data, the SAS team discovered many missing values in the KDD data. This white paper shows how the SAS team used SAS Enterprise Miner with the 2009 KDD data to build several kinds of predictive models and determine the most accurate one.
Sponsored by SAS
Individual, Student, and Team memberships available.