Determining Transportation Mode through Cellphone Sensor Fusion

Contextual awareness is a field that allows computer systems to provide users and their applications with valuable information about the world around us, such as the ‘Mode of Transportation’ of a given user. This data can be used in many different ways, for example, the analysis of traffic patterns and CO2 emissions. In this thesis, I designed and implemented a mobile application that determines the user’s current mode of transportation. This is done by applying a Boosted Naive Bayes classifier to data collected from the sensors of the mobile device. Generating this Boosted Naive Bayes classifier required me to first develop an application to collect training data. I then applied several Naive Bayes classifiers to these data. Finally, I applied the AdaBoost algorithm to these classifiers to obtain the boosted classifier. The Boosted Naive Bayes classifier performs better than its unboosted counterparts. Boosting can be a useful tool to apply to context recognition problems. Thesis Supervisor: Professor Seth Teller Title: Electrical Engineering and Computer Science Department