Clustering of Time Series Data for Enhanced Forecasting: A Comparative Study and Practical Applications

Time series forecasting plays a pivotal role in various domains, such as finance, healthcare, and supply chain management. Traditional forecasting methods often assume that all time series follow a similar pattern, which may not hold true in real-world scenarios. Clustering time series data has emerged as a promising approach to address this limitation, enabling the identification of groups of time series that exhibit similar patterns as well as improving forecast predictions and computational time and resourced needed. In this thesis several approaches from the literature are compared as well as some more advanced techniques for clustering of time series with a particular attention when the dimension of the dataset increase.