Repeatability index: A functional metric assessing intra-regime variability in helicopters

The assessment of intra-class variability is a challenging task in the context of multivariate time-series classification, as it hinders algorithms from effectively identifying characteristic patterns, thereby jeopardizing overall performance. To tackle this issue, we propose a novel indicator that we call repeatability index. Rooted in a functional data framework, this index effectively quantifies variability among instances belonging to the class while accounting for temporal dependencies. Besides, by combining shape and duration variability indicators, the repeatability index is specifically suitable for assessing intra-class variability in multivariate time-series characterized by intrinsically different durations. We prove the effectiveness and versatility of the proposed indicator in the context of time-series classification for Flight Condition Recognition (FCR). FCR algorithms aim at classifying executed maneuvers from time-series data describing the dynamics of the helicopter ...