Abstract

A large number of variables and multivariate analyses are required to study complexity. Network analysis is also a valuable complex systems analysis tool. Recently techniques were developed where correlation matrixes are presented as networks, known as correlation graphs. These have been used mainly to study structures in time series data, often in the financial stock markets, and correlation graphs have not been widely adopted. The aim of this paper is to demonstrate correlation graphs as a general research tool by using intuitively understandable data. Correlation graphs of human body shape dimensions were generated and noise reducing techniques are demonstrated. The results support the underlying structures of human biology and gender is distinguished with an accuracy of 89%. Weight plays a vital role and height is more influential in males than in females. With age, males and females become more alike. The potential application of the correlation graphing methodology appears extensive.

Key words: Correlation graph, asset graph, multivariate analyses, graph theory, body shape.