Title

Classification Trees with Synthetic Features for Multi-class Problems

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Mathematics

Date of Award

Spring 2019

Abstract

Using synthetic features for classification is one of the most important fields of study in statistics. This thesis presents the significance of synthetic features for multi-class problems with different types of data set. The purpose was to compare the misclassification error between using all the synthetic features and none of them by using software R. Original code was written for each individual feature and the method of finding probabilistic misclassification error. The data set is either randomly generated or given by the professor. The idea was to use boosting, stump, and probabilistic models to determine the misclassification error. Experiments were done with different settings for each data set. From the experiments, all the results indicated that synthetic features helped to reduce the misclassification error and different synthetic feature worked for different types of data.

Advisor

Thomas R. Boucher

Subject Categories

Mathematics | Physical Sciences and Mathematics

Share

COinS