CLUSTER ANALYSIS AND PRINCIPAL COMPONENTS
ANALYSIS IN THE EVALUATION OF MULTIVARIATE TABLES (N X P).
Paula Roberto CURI[1]
Alexandre ALVES2
Patrick Johannes BECKERS[2]
Lúcia TERADA[3]
§ ABSTRACT: The situation of the
countries is studied based on a collection of basic, health, economic and
cultural indicators published by UNICEF in 1990. It was used multivariate
statistical methods: Principal Components Analysis and Cluster Analysis. The
study was performed to countries of all the continents, with 36 countries whose
choice criterion was to include countries with complete information (to the 16
used variables) and of the most varied degrees of development. The first
principal component was interpreted as a measure of the "development
degree "which made it possible the order of the countries. The most
discriminating variables were the childish death rates and death of children
with less than 5 years old and the expectation of living.
§ KEYWORDS: Principal components
analysis; Cluster analysis; multivariate tables.