The DDG-classifier in the functional setting

• J. A. Cuesta-Albertos ^[1] ; M. Febrero-Bande ^[2] ; M. Oviedo de la Fuente ^[2]
  1. [1] Universidad de Cantabria

     Universidad de Cantabria

     Santander, España

     
  2. [2] Universidade de Santiago de Compostela

     Universidade de Santiago de Compostela

     Santiago de Compostela, España

     
• Localización: Test: An Official Journal of the Spanish Society of Statistics and Operations Research, ISSN-e 1863-8260, ISSN 1133-0686, Vol. 26, Nº. 1, 2017, págs. 119-142
• Idioma: inglés
• DOI: 10.1007/s11749-016-0502-6
• Texto completo no disponible (Saber más ...)
• Resumen

  • The maximum depth classifier was the first attempt to use data depths instead of multivariate raw data in classification problems. Recently, the DD-classifier has addressed some of the serious limitations of this classifier but issues still remain. This paper aims to extend the DD-classifier as follows: first, by enabling it to handle more than two groups; second, by applying regular classification methods (such as kNN, linear or quadratic classifiers, recursive partitioning, etc) to DD-plots, which is particularly useful, because it gives insights based on the diagnostics of these methods; and third, by integrating various sources of information (data depths, multivariate functional data, etc) in the classification procedure in a unified way. This paper also proposes an enhanced revision of several functional data depths and it provides a simulation study and applications to some real data sets.