Thesis Title

Applied Multivariate Statistical Methods

Date of Graduation

Spring 2005

Degree

Master of Science in Mathematics

Department

Mathematics

Committee Chair

Yingcai Su

Keywords

random vectors, random matrix, classification, dissimilarity matrix, cluster

Subject Categories

Mathematics

Abstract

An understanding of multivariate statistical methods is essential to many fields of study. There are five principle multivariate methods discussed in the thesis. The first is principle components analysis, which relies heavily on the variance of the observations, and is a starting place for many of the other methods discussed herein. Factor analysis is a controversial method whereby the analyst can draw conclusions about underlying factor common to all the observations. A common use for factor analysis is in drawing similarities between different objects such as countries or species of animals. Classification includes several different procedures whereby the data can be grouped into known groups of like properties. Clustering, a method very similar to classification, attempts to create clusters of like observations and allocate objects into those clusters. Multidimensional scaling is a method used to interpret a graph of observations into an easily perceived two or three dimensional plot. Multidimensional scaling has many applications in the medical and technological fields.

Copyright

© Aaron Nieuwsma

Citation-only

Dissertation/Thesis

Share

COinS