Iterative data analysis for sensing applications

Smartphones and other sensor devices can generate large amounts of data in a short time. Even if their computing capacity and battery lifetime can be limited, they usually have good communication capabilities, so that they can take advantage of remote services. Large-scale data analysis offers methodology, which can be used to improve functionality of the applications and extend the user activity. This creates a need for an iterative data analysis system, which offers streaming processing, data flow management, and machine learning algorithms suitable for complex sensing data. My PhD work aims to design principles and practices for an iterative data analysis algorithms and workflow. As a case study, we are building a mobile application that measures context factors' combined impact to energy consumption. Our approach will be useful for different types of cases, where it is important to understand complex data sources in real time. My work is supervised by professor Sasu Tarkoma and made in collaboration with Dr Eemil Lagerspetz and Dr Petteri Nurmi at the University of Helsinki.