Change point detection for compositional multivariate data

Change point detection algorithms have numerous applications in areas of medical condition monitoring, fault industrial processes, human activity analysis, climate change detection, and speech recognition. We consider the problem on compositional multivariate data (each sample is a probability mass function), which practically important sub-class general data. While change-point well studied univariate setting, there are few viable implementations for data, existing methods do not perform In this paper, we propose parametric approach Moreover, using simple transformations extend our to handle any Experimentally, show that method performs significantly better competitive compared available state art implementations.