APPLICATION OF MATRIX PROFILE TECHNIQUES TO DETECT INSIGHTFUL DISCORDS IN CLIMATE DATA

<p>The definition and extraction of actionable anomalous discords, i.e. pattern outliers, is a challenging<br>problem in data analysis. It raises the crucial issue of identifying criteria that would render a discord<br>more insightful than another one. In this paper, we propose an approach to address this by<br>introducing the concept of prominent discord. The core idea behind this new concept is to identify<br>dependencies among discords of varying lengths. How can we identify a discord that would be<br>prominent? We propose an ordering relation, that ranks discords, and we seek a set of prominent<br>discords with respect to this ordering. Our contributions are threefold 1) a formal definition,<br>ordering relation and methods to derive prominent discords based on Matrix Profile techniques,2)<br>their evaluation over large contextual climate data, covering 110 years of monthly data, and 3) a<br>comparison of an exact method based on STOMP and an approximate approach that is based on<br>SCRIMP++ to compute the prominent discords and study the tradeoff optimality/CPU. The<br>approach is generic and its pertinence shown over historical climate data.</p>