Application of Cluster Analysis to Aerometric Data

Abstract

The NORMIX data analysis program, which incorporates cluster analysis and multivariate statistical analysis routines, was modified and revised for use In a UNI VAC 1110 computer. The revised program was tested on three sample data sets and produced results in agreement with those from the original program. The NORMIX program was then used to evaluate and analyze eight sets of aerometric data from various sources. Comparison of the performance of NORMIX with two other cluster analysis algorithms, MIKCA and SAS CLUSTER, revealed that all three programs produce similar results In terms of hierarchical clustering, but NORMIX produces considerably more statistical evaluation and Information to the user. Thus NORMIX is recommended as the most useful cluster analysis program of these three.