Multiclass classification of functional data with repeated measures: an application to gait data

M. C. Aguilera Morillo, A. M. Aguilera del Pino, M. Escabias Machuca

In this work we are focused on the multiclass classification of gait data according to different experimental conditions to which the subject is exposed. Gait data is a particular case of kinematic data usually collected in the format of linear acceleration or angular rotation. Linear acceleration and angular rotation represent the trajectory of the human movement, which can be identified as a function (varying on a continuum). On the other hand, kinematic studies are usually characterized by the presence of repeated measures. Then, considering the continuous nature of gait data, and the presence of repeated measures, new methods are proposed in the framework of functional data analysis to solve the multiclass classification problem of gait data.

Palabras clave: Gait analysis functional classification roughness penalty

M. Febrero Bande, J. Álvarez Liébana, M. Oviedo de la Fuente

J. A. Cuesta Albertos, S. Dutta

C. Ramos Carreño, A. Suárez González, J. L. Torrecilla Noguerales