Three Studies on the Practical Use of Machine Learning Techniques in Analytics

Loading...
Thumbnail Image

Authors

Ying, Cecilia

Date

2025-05-05

Type

thesis

Language

eng

Keyword

Text Analytics , Machine Learning , Large Language Models , Automated Credit Assessment , ML Model Bias and Model Correction

Research Projects

Organizational Units

Journal Issue

Alternative Title

Abstract

The thesis investigates the application and societal implications of advanced machine learning techniques in data analytics through three empirical studies. The first study evaluates large language models (LLMs) in conversation analytics, finding that LLM-based configurations significantly outperform traditional methods in topic extraction tasks. The second study examines the impact of data sparsity on automated consumer credit assessments using machine learning algorithms, highlighting risks to borrowers who lack a substantial financial history. The third study introduces a post-processing technique, Subgroup Threshold Optimization (STO), to address performance disparities identified in the second study. This technique strikes a balance between fairness and performance, providing insights into the trade-offs between profit and fairness. The research aims to guide the responsible implementation of machine learning technologies, providing best-practice guidelines for practitioners.

Description

Citation

Publisher

License

Queen's University's Thesis/Dissertation Non-Exclusive License for Deposit to QSpace and Library and Archives Canada
Intellectual Property Guidelines at Queen's University
Copying and Preserving Your Thesis
This publication is made available by the authority of the copyright owner solely for the purpose of private study and research and may not be copied or reproduced except as permitted by the copyright laws without written authority from the copyright owner.
Attribution-NonCommercial-NoDerivatives 4.0 International

Journal

Volume

Issue

PubMed ID

External DOI

ISSN

EISSN