Cophenetic-based fuzzy clustering of time series by linear dependency

In this work, a new approach to cluster large sets of time series is presented. The proposed methodology takes into account the dependency among obtain fuzzy partition set observations. A two-step procedure accomplish First, cophenetic distances, based on linear cross-dependency measure, are obtained. Second, these distances used as an input non-Euclidean relational clustering algorithm. As result, we robust capable detecting groups with different types cross-dependency. We illustrate usefulness stated through some Monte Carlo experiments and real data example. Our results show that in work substantially improves hard partitioning alternative.