Nonmetric clustering: new approaches for ecological data

Abstract

Ecological studies and multispecies ecotoxicological tests are based on the examination of a variety of physical, chemical and biological data with the intent of finding patterns in their changing relationships over time. The data sets resulting from such studies are often noisy, incomplete, and difficult to envision. We have developed machine learning and visualization software to aid in the analysis, modelling, and understanding of such systems. The software is based on nonmetric conceptual clustering, which attempts to analyze the data into clusters that are strongly associated with several measured parameters. Our analysis and visualization tools not only confirmed suspected ecological patterns, but revealed aspects of the data that were unnoticed by ecologists using conventional statistical techniques.<>