A1402
Title: Clustering time series of counts
Authors: Luis Sousa - University of Aveiro (Portugal) [presenting]
Isabel Pereira - University of Aveiro (Portugal)
Magda Monteiro - University of Aveiro (Portugal)
Abstract: The clustering of time series has proven to be of interest in various fields, ranging from economics and finance to environment and medicine, among others. The objective is to group similar items according to a criterion suitable for the problem. Specifically, this aims to help outline strategies for better decision-making in the context of logistics, particularly in maritime ports. Some of the problems that may arise within this context range from predicting the number of ships arriving at one port to the clustering of ships based on the types of materials they are transporting. Much of the work developed over the recent decades has been conducted within the framework of continuous-valued time series, with few studies on clustering for count time series. The aim is to establish and apply model-based clustering to appropriately define discrete-valued time series, particularly those that allow for overdispersion and/or zero inflation. The idea is to use a finite mixture model that accommodates the mentioned characteristics, and several existing techniques, such as the selection of the number of clusters, estimation using expectation-maximization and model selection, are applicable. The methodology proposed employs a mixture of count models to cluster discrete-valued time series, in which each time series is allocated to a specific process. A simulation study is carried out, and an illustration with a real data set is made as well.
