Hidden semi-Markov chains : a new tool for analyzing nonstationary discrete sequences

Spatial structure in plant architectures can be described as discrete sequences which are very often nonstationary. We propose to use hidden semi-Markov chains for analyzing samples of such sequences. Having chosen a family of hidden semi-Markov chains, it is possible to estimate parameters, check for goodness of fit of the data and then use the fitted model to enhance our understanding of the biological machanism underlying the establishment of the measured spatial sequences. In the process of model building, usually decomposed into stages of specification, inference and validation, the roles of the first and last stages are then strengthened. Characteristic distributions can be computed from the hidden semi-Markov chain parameters while their empirical equivalents extracted from a sample of sequences constitute a set of exploratory tools. This methodology is illustrated by the analysis of the branching structure of the first annual shoot of apple tree trunks. The first annual shoot is the portion of the trunk established during the first year of growth and its branching structure after two years is assumed to be a good predictor of the adult structure of the tree. After an exploratory analysis, the author chose to estimate a left-right hidden semi-Markov chain composed of six successive transient states followed by a final absorbing state. This methodology can be applied to the detailed comparison of the branching structure of apple cultivars

Saved in:
Bibliographic Details
Main Author: Guédon, Yann
Format: conference_item biblioteca
Language:eng
Published: UTC
Subjects:U10 - Informatique, mathématiques et statistiques, F50 - Anatomie et morphologie des plantes, Malus, port de la plante, modèle de simulation, ramification, méthode statistique, modélisation, http://aims.fao.org/aos/agrovoc/c_4553, http://aims.fao.org/aos/agrovoc/c_5969, http://aims.fao.org/aos/agrovoc/c_24242, http://aims.fao.org/aos/agrovoc/c_1057, http://aims.fao.org/aos/agrovoc/c_7377, http://aims.fao.org/aos/agrovoc/c_230ab86c,
Online Access:http://agritrop.cirad.fr/390693/
Tags: Add Tag
No Tags, Be the first to tag this record!