A0718
Title: Time series entropy: Clustering for decision-making
Authors: Miguel Angel Ruiz Reina - Universidad de Malaga (Spain) [presenting]
Abstract: The multidimensional classification generates information gaps for researchers, practitioners, or businesses; Information Theory clustering is a solution to understanding seasonal decision-making time series. We propose an automatic clustering system based on the multi-optional based on Shannon entropy, combining techniques that relax the usual assumptions of statistics to understand the data set. Clustering metric methods are crucial for many real-world applications, and distance metrics provide learning models with better performance than is generally achieved. This unsupervised classification method automatically adjusts spatio-temporal observations and organises the data set for decision-making. The empirical field of application is tourist accommodation decision-making among hotels, tourist apartments, campsites and rural apartments for foreign tourists visiting Spain from January 2001 to January 2022. The intracluster verification criterion confirms the similarity of the members of the group. In this way, policymakers could adjust their offers or impact policies based on the seasonal typology studied. It is possible to convert time series from High-Dimensional Time Series to Reduction-Dimensional Time Series by recognising behaviour patterns of foreign tourist demand in Spain. This statistical learning model allows for building analysis models on large volumes of data and providing unsuspected knowledge in the initial exploratory analysis of the data.
