Date of Award

8-27-2018

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Department of Mathematics and Statistics

First Advisor

Ryan A. Kappedal, PhD.

Abstract

Phenotype classification has become an increasingly important genomic research method for disease identification and treatment. Phenotype classification is the investigation into the genetic information concerned with locating biomarkers (features) in order to identify an observed effect. The primary challenge associated with phenotype classification is with analyzing the data due to the inherent high-dimensionality of DNA data. As a result, phenotype classification faces challenges with feature selection, and consequently, classification accuracy. This research developed a methodology to alleviate these challenges while improving classification performance. The methodology leverages concepts of compressive sampling, to arrive at a process that identifies features most relevant to the phenotype. Additionally, this research presents a probabilistic acceptance of the RIP and uses it to qualify dataframes constructed by the proposed methodology. Overall, I found this methodology as a viable approach to dimension reduction and feature selection, which improved phenotype classification accuracy.

AFIT Designator

AFIT-ENC-DS-18-S-001

DTIC Accession Number

AD1063251

Share

COinS