Telecommunication has a common problem of suffering from high churn rate, e.g. an existing customer may leave the service provider or shift to another. Churn prediction is the task of predicting the likelihood that a certain group of customers churn, and customer segmentation has been widely applied for a target marketing strategy. Previous work has adopted various clustering and classification methods, among which support vector machines and neural networks have achieved better performance, but the possible reason why customers churn is rarely explored in details. This article aims to address the gap of understanding customer churn using customers’ usage of services. To do this, an integrated framework containing two phases was built: we carried out predictive analysis using churn prediction and association analysis (phase one), based on which we provided recommendations in the decision analysis (phase two). The experimental results showed that the proposed Usage-based Decision Tree (UsageDT) achieved better performance in understanding the churners/non-churners in terms of interpretability, higher precision, and much higher true positive rate. The usage-based association analysis was integrated seamlessly with churn prediction, and deepened our understanding of the churn phenomena. Evaluation carried out with two professionals in telecommunication industry also revealed the usefulness of our recommendations.

Exergaming as a tool to combat obesity yields an interesting take on the problem of design and implementation of activity recognition systems for truly mobile games that achieve moderate levels of intensity. This work presents SoccAR, a mobile, sensor-based wearable exergaming system with fine-grain activity recognition. The system in this paper presents a recognition algorithm for the appropriate classification of 26 movements by extracting a large number of features and selecting the most important, as well as developing a multiple model strategy to better classify movements. This movement strategy allows for a trade off of detailed classification versus classification speed. A metric to define the accuracy in terms of the importance of particular movements is defined. The scheme presented develops a framework for more accurately classifying movements with a smaller number of features for a large, multiclass real-time environment. This results in a more accurate classification of movements, with an F-score in cross-validation of .937 using a PUK-kernel based SVM and multiple models, to .755 using only a single RBF-based model and 20 features.