Classifying Mobile-Phone Users With An Information Theory Approach ∗

In this paper, we use a learning method based on information theory to classify mobilephone users as 2G and 3G customers. In our method, we aim at looking for informative and discriminatory feature subset. Then, we build classification models with these feature subsets. We find some general properties of promising 3G users, i.e., the false positive predictions. These properties should be useful for easily differentiating the 3G mobile phone users.