This paper proposes a Host-based Intrusion Detection System (HIDS) for Android mobile devices that employs Machine Learning (ML) algorithms to identify probability of intrusion by analysing the system log files [1], [2]. Firstly, we identified the list of features that characterize effectively the impact of malware on the mobile device and maximize the effectiveness of the ML techniques for intrusion detection. For example, a high consumption of computation or communication resources (e.g., CPU, memory, battery, network bandwidth) may indicate a malicious activity due to malware running on the device.