Bayesian logistic regression: An application for carbonisation damage in four wood species. Issue 1 (2nd January 2023)

• Record Type: Journal Article
• Authors: Vasconcelos, Juliano S.; Vasconcelos, Julio C. S.; dos Santos, Denize P.; Villegas, Cristian; De Araujo, Victor A.; Biaggioni, Marco A. M.
• Is Part Of: Wood material science & engineering. Volume 18:Issue 1(2023)
• Page Start: 107
• Page End: 111
• Publication Date: 2023-01-02
• Subjects: Binary data -- credibility intervals -- forest species -- Markov chain -- statistical analysis; Wood -- Periodicals; Wood products -- Periodicals; Building, Wooden -- Periodicals; 624.184
• DOI: 10.1080/17480272.2021.1992649
• Languages: English
• ISSNs: 1748-0272

Abstract

ABSTRACT: This paper aimed to measure the carbonization action in four wood species with different densities formed by a variety with global utilization, eucalyptus grandis, and other three native species from Brazil, cupiuba, guapuruvu and morototoni. The analysis included the verification of damages from carbonization at high temperature as well as the influence of these covariates by the Bayesian Logistic Regression. The analysis verified the convergences from R-Hat diagnostic and Markov chains. Statistical model of temperature-species covariates was selected, since it presents the possibility of a more plausible explanation for the assessed data set. Eucalyptus grandis and guapuruvu wood species showed the greatest damage volumes caused by carbonization. Density was not a covariate that interfered, statistically, in this analysis of high temperature damages. Regarding wood species, the eucalyptus grandis differed significantly from the Bayesian model, both from cupiuba and morototoni woods. The detailed analysis evinced the efficient performance of the Bayesian model, which is an alternative option for studies about the carbonization of forest resources.