DYNAMICS OF SHEEP PRODUCTION IN BRAZIL USING PRINCIPAL COMPONENTS AND AUTO-ORGANIZATION FEATURES MAPS

Abstract

The present study aims to evaluate the multitemporal dynamics of sheep production in Brazil, considering the Federal Government's official data from 1976 to 2010. Principal Component Analysis (PCA) and Self-Organizing Maps (SOFM) were used. PCA reduced data redundancy and the colour composition of the first three principal components evidenced temporal patterns. In the SOFM classification, a type of Artificial Neural Network (ANN) with non-supervised training, different dimensions of the Kohonen map were tested. The results obtained by the two methods are complementary, evidencing the development of sheep production in the country. The southern and northeast maintained the tradition of sheep production over the analyzed period. The municipalities in the Midwest showed production growth. The techniques used were effective in multitemporal analysis, providing a greater understanding of the dynamics of sheep production in Brazil and providing subsidies for the development of appropriate public policies for its expansion.