Banca de QUALIFICAÇÃO: MARCELA PORTELA SANTOS DE FIGUEIREDO

Uma banca de QUALIFICAÇÃO de DOUTORADO foi cadastrada pelo programa.
STUDENT : MARCELA PORTELA SANTOS DE FIGUEIREDO
DATE: 17/08/2022
TIME: 15:00
LOCAL: https://meet.google.com/nsb-jkgi-cqd
TITLE:

The generalized composition that generates growth/degrowth models applied to agricultural sciences

KEY WORDS:

The dual-pool models; Animal growth; In vitro gas production technique; Model composition.

PAGES: 75
BIG AREA: Ciências Agrárias
AREA: Agronomia
SUMMARY:

The aims of this work were: To generalize the method of Santos et al (2019), add the operation of composition of functions to the method, and create new models applied to data from the agricultural sciences. The first chapter is a literature review related to growth models applied to animal growth and to the kinetics of gases produced by the in vitro technique. In the second chapter, a new model was presented, called the Mitscherlich-Logistic Bicompartmental (MLB), obtained by the method of Santos et al. (2019), applied to corn and sunflower silage data and their mixtures and this is the first article of the thesis. The database was taken from the article by Leite (2006). The proposed model was compared to the Logistic Bicompartmental (LB), Von Bertalanffy-Gompertz Bicompartmental (VGB), and Logistic-Von Bertalanffy Bicompartmental (LVB) models. The estimation of the parameters was performed using the least squares method of the R^™ software, using the nls function that uses the Gauss-Newton algorithm. The proposed model obtained superior results concerning the other models referring to the diets SM - corn silage and 340SG - 660 g kg-1 of corn silage and 340 g kg-1 of sunflower silage. The quality of fit evaluators used were: Adjusted Coefficient of Determination R_(aj.)^2, The Resicual Mean Squares (RMS), Akaike Information Criterion (AIC) and Mean Absolute Deviation of residues (MAD). The gas production curves show differences in the diets when analyzed through the curve identity test. In the third chapter of the thesis, the generalization of the method by Santos et al. (2019), adding the function composition operation. A new unicompartmental model was created by combining the Brody model with that of Von Bertalanffy, which was applied to animal growth data from California rabbits. The data used in this article  describes the observations obtained in the experiment by  Curi, Nunes, and Curi (1986). The estimation of the parameters was also made by the method of least squares of the software R®, through the function nls that uses the algorithm of Gauss-Newton. The proposed model was compared to the following models: Logistic, Brody, Von Bertalanffy, Gompertz, Richards, through the quality of fit criteria the adjusted coefficient of determination R_(aj.)^2, The residual mean squares (RMS), Akaike Information Criterion (AIC) and Mean Absolute Deviation of the residues (MAD) and cluster analysis was performed to define the best model. The Proposed model obtained results equivalent to the Gompertz and Richards models. These three models were the most adequate to fit the California rabbit data. Therefore, this work contributed, by adding the composition operation. In addition, the model generated by this method is effective to fit data from the California breed, this chapter is the second article of the thesis.

BANKING MEMBERS:
Externa à Instituição - MERCIA REGINA PEREIRA DE FIGUEIREDO
Interno - 2279488 - ANTONIO SAMUEL ALVES DA SILVA
Externo à Instituição - CICERO CARLOS RAMOS DE BRITO - IFPE
Interno - 1756995 - FRANK SINATRA GOMES DA SILVA
Presidente - 1950420 - GUILHERME ROCHA MOREIRA