Mobile malware is on the rise. Indeed, due to their popularity, smartphones represent an attractive target for cybercriminals, especially because of private user data, as these devices incorporate a lot of sensitive information about users, even more than a personal computer. As a matter of fact, besides personal information such as documents, accounts, passwords, and contacts, smartphone sensors centralise other sensitive data including user location and physical activities. In this paper, we study the problem of malware detection in smartphones, relying on supervised-machine-learning models and big-data analytics frameworks. Using the SherLock dataset, a large, publicly available dataset for smartphone-data analysis, we train and benchmark tree-based models to identify running applications and to detect malware activity. We verify their accuracy, and initial results suggest that decision trees are capable of identifying running apps and malware activity with high accuracy.