A0884
Title: On identifying functional motifs and using them to aid forecasting and missing value imputation
Authors: Jacopo Di Iorio - Emory University (United States) [presenting]
Francesca Chiaromonte - The Pennsylvania State University (United States)
Marzia Cremona - Universite Laval (Canada)
Abstract: Functional data analysis faces a novel challenge: Identifying functional motifs, or shapes, that may be repeated multiple times within each functional observation or across multiple curves belonging to the same set. To address this issue, funBIalign is introduced, a multi-step approach that employs agglomerative hierarchical clustering with complete linkage and functional distances based on mean squared residue scores and virtual error. These distances enable funBIalign to detect functional motifs that may be shifted or scaled along the y-axis. To validate the effectiveness of the methodology, simulations and case studies that demonstrate its ability to identify functional motifs are presented. The identification of functional motifs can be leveraged to solve other important problems in functional data analysis. For instance, portions characterized by the same motif are hypothesized to evolve similarly, which can aid in forecasting and the missing portions imputation problems.
