Curve clustering based on polynomial splines and nonsmooth optimisation

This paper examines clustering in datasets where each record is presented as a curve (trend). Our clustering approach is based on nonsmooth and nonconvex optimisation, namely the clustering problems have been formulated as mathematical programming problems and several optimisation methods from GANSO optimisation library have been applied to solve the problems. The least squares and uniform optimisation criteria have been examined. The numerical experiments have been carried out on the Pezzack dataset. The obtained results encourage the continuation of the research.