International Day Of Women in Statistics and Data Science

Functional data analysis has attracted considerable attention in recent years, and its applications appear in physical processes, genetics, biology, meteorology, and signal processing. Many modern applications produce data best viewed as functions rather than finite-dimensional vectors because of their nature. Beyond the challenges of collecting and preprocessing such data, efficiently handling large volumes of functional observations has become an urgent concern. On one hand, functional data takes values in an infinite-dimensional space, which is challenging to handle with classical methods. On the other hand, ignoring the functionality aspect of data will lead to information loss. In this talk, we highlight model-based clustering methods, a powerful tool in machine learning for identifying subgroup-specific patterns. Specifically, we introduce new model-based clustering techniques for functional data, regardless of the Gaussian assumption. The performance of each algorithm is evaluated through simulations and real-world datasets, and the results confirm their efficiency.