O. Ojo, R. E. Lillo Rodriguez, A. Fernández Anta
We present theoretical definitions and properties of the FastMUOD indices, used for outlier detection in functional data. FastMUOD detects outliers by computing for each curve: amplitude, magnitude and shape indices meant to target the respective types of outliers. We also provide sample versions of these indices and prove their convergence. We also propose some ideas for adapting FastMUOD indices for outlier detection in multivariate functional data. We test some of these ideas on various simulated and real datasets.
Keywords: outlier detection, functional data, FastMUOD
Scheduled
GT06 Functional Data Analysis II. Tools and apps
June 8, 2022 4:00 PM
Grade Hall