Title: Dynamic time warping-based fuzzy clustering with spatial information
Authors: Pierpaolo Durso - University of Rome Sapienza (Italy)
Marta Disegna - Faculty of Management, Bournemouth University (United Kingdom) [presenting]
Riccardo Massari - Sapienza University (Italy)
Abstract: Clustering of spatial multivariate time series finds application in several fields, such as economics, marketing, finance, and medicine. Classification of such complex data requires to take into account both spatial and time dimensions. In particular, one should consider (1) the spatial nature of the units to be clustered, (2) the characteristics of the space of multivariate time trajectories, and (3) the uncertainty related to the assignment of a geographical unit to a given cluster on the basis of the above complex features. Existing spatial-time clustering models can be distinguished into non-spatial time series clustering based on a spatial dissimilarity measure; spatially constrained time series clustering; density-based clustering; model-based clustering. The aim is to discuss a novel spatially constrained multivariate time series clustering for units characterised by different levels of spatial contiguity. In particular, the Fuzzy Partitioning Around Medoids algorithm with Dynamic Time Warping distance and spatial penalization terms is applied to classify multivariate time series. This method has been theoretically presented together with the illustration of the empirical case study related to the identification of touristic agglomerations of cities and towns belonging to the same macro destination and characterised by similar touristic flows over time.