J. J. Egozcue, V. Pawlowsky-Glahn
The initial probabilities of incompatible events form a composition. The likelihood is a vector of probabilities of the observed result conditional to the considered events and it can be considered as a composition. The group operation for compositions, the perturbation, consist of componentwise multiplication of initial and likelihood probabilities. The Bayes' formula is the compositional perturbation of initial probabilities and the likelihood. The Bayes' formula can be taken as the paradigm of information assimilation, as the three compositions involved are the initial, the input and the final evidences about the actual event. The perturbation of evidence compositions is a vectorial addition. Therefore, evidence information has a direction and a modulus or scalar magnitude. The Aitchison norm for compositions provide a measurement of evidence information and its properties allow considering e-information as a proper scalar measurement of information.
Palabras clave / Keywords: Bayes, information, composition, Aitchison norm, evidence, perturbation
Programado
Sesión GT02-3: Análisis Multivariante y Clasificación (AMyC-3). Organizadora: Eva Boj del Val
29 de mayo de 2018 15:20
Sala 5
