The objective of this project was to investigate and compare several existing methods for clustering and classifying data as well as some of my own design. Under the employment of BAE Systems, Inc., I tested algorithms designed to identify clusters and determine parameter estimates from continuous data that were known to be in the form of Gaussian mixtures. I also looked into methods for classifying discrete data that were known to originate from multiple unique sources, in hopes of being able to categorize these data automatically based on previous knowledge. Through my findings, I was able to determine which algorithm performed the best. I also found multiple viable methods for automatically categorizing the discrete data that BAE Systems was interested in.
Worcester Polytechnic Institute
Major Qualifying Project
All authors have granted to WPI a nonexclusive royalty-free license to distribute copies of the work, subject to other agreements. Copyright is held by the author or authors, with all rights reserved, unless otherwise noted.