Online Program

Predicting Buying Behavior: IT Software Customer Clustering with R and Weka (303047)

Keywords: clustering, data mining, R, Weka, K-means, DBSCN

The proposed presentation centers on using clustering, a data mining technique, to group customers of an IT software company by shared characteristics and behavior to create a data-driven customer relationship management system (CRM) based on likely buying behavior and patterns. The following variables of interest will be covered: time to purchase, products purchased, method of purchase (online, phone, etc.), size of software license, free downloads prior to purchase, and additional online behaviors. Several types of algorithms—including partitioning-based and density-based—will be compared and evaluated on the efficiency, appropriateness, and effectiveness of handling the data problem in question. The goal is to determine not only the best algorithmic method for customer clustering, but also to evaluate two of the most commonly used free software packages for data mining. A select number of clustering algorithms will be executed in both R and Weka, and then within R with the Weka connection.The presentation will conclude with suggestions for how to use statistical methods to verify the clusters produced.