Change Detection in Multivariate Datastreams Controlling False Alarms

We introduce QuantTree Exponentially Weighted Moving Average (QT-EWMA), a novel change-detection algorithm for multivariate datastreams that can operate in nonparametric and online manner. QT-EWMA be configured to yield target Run Length (ARL\(_0\)), thus controlling the expected time before false alarm. Control over alarms has many practical implications is rarely guaranteed by algorithms monitor whose distribution unknown. Our experiments, performed on synthetic real-world datasets, demonstrate controls ARL\(_0\) alarm rate better than state-of-the-art methods operating similar conditions, achieving comparable detection delays.