Interpreting Intermittent Bugs in Mozilla Applications Using Change Angle

David Tse Jung Huang, Yun Sing Koh, Gillian Dobbie

Abstract: Learning in evolving environments involves learning from data where the statistical characteristics can change over time. Current change detection algorithms that are used online for data streams detect whether a change has occurred in the data but there is always a detection delay. None of the existing online techniques can accurately pin-point the exact location of when the change starts to occur, which can be critical. We present a novel method Change Angle and we show, for the first time, how to pin-point online the location at which change starts to occur. We apply our Change Angle method in the application area of software revision control using Mozilla data, where it is important to detect not only the presence of change but also to pin-point accurately the location of when change starts to occur.