COMING SOON! PQDT Open is getting a new home!

ProQuest Open Access Dissertations & Theses will remain freely available as part of a new and enhanced search experience at

Questions? Please refer to this FAQ.

Dissertation/Thesis Abstract

Tree-Based Ensemble Classification Algorithms for Genomic Data
by Doan, Quoc A., M.S., California State University, Long Beach, 2020, 54; 27736947
Abstract (Summary)

Random Forest is one of the widely used tree-based ensemble classification algorithm. Many aspects of building tree ensembles are introduced to reduce correlation among decision trees within the forest. Bootstrap is used in Random Forest to reduce bias decision tree and to decide split in every decision tree. Classification by Ensembles from Random Partitions (CERP) is a different algorithm to create an ensemble. CERP randomly partition the data instead of using bootstrap and create multiple ensembles instead of one. A forest consists of several decision trees, an ensemble of trees. While Random Forest builds a forest, CERP builds an ensemble of forests. A base classifier in Random Forest uses an exhaustive search to find a split. On the other hand, the Generalized, Unbiased, Interaction Detection and Estimation (GUIDE) algorithm uses statistical hypothesis testing which is faster than exhaustively search algorithm and is able to detect interaction using a statistical method. This thesis investigated tree-based ensemble classification algorithms that include the CERP and GUIDE and Random Forest for genetic data.

Indexing (document details)
Advisor: Moon, Hojin
Commitee: Suaray, Kagba, Zhou, Tianni
School: California State University, Long Beach
Department: Mathematics and Statistics
School Location: United States -- California
Source: MAI 81/11(E), Masters Abstracts International
Subjects: Statistics, Medicine, Mathematics
Keywords: Data science, Decision trees, Ensemble, Machine learning, Random forest, Statistical learning
Publication Number: 27736947
ISBN: 9798645443122
Copyright © 2021 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy