Date of Publication

3-19-2025

Document Type

Master's Thesis

Degree Name

Master of Science in Mathematics

Subject Categories

Mathematics

College

College of Science

Department/Unit

Mathematics and Statistics Department

Thesis Advisor

Angelyn R. Lao

Defense Panel Chair

Jose Tristan Reyes

Defense Panel Member

Kristine Joy E. Carpio
Fatimah Abdul Razak

Abstract/Summary

Methods in machine learning have been used in recent decades to aid market participants in determining the future direction of stock markets, which is imperative for any investment decision to yield high financial returns and minimize risks. Several studies have integrated persistent homology into machine learning, and it has been shown that this approach improves accuracy in inferencing imaging datasets, recognizing patterns and predicting time series data. In computational topology, persistent homology is a tool that keeps track of data features that persist across different scales. Application of persistent homology obtains invariant topological features which may be used as input data for machine learning models. In this study, we choose indices from the Philippine Stock Exchange as our data for prediction: the Composite Index, the Service Index and the Industrial Sector Index. The stock returns, technical indicators and topological features obtained from the historical data are used in the machine learning models, artificial neural network and support vector machine. We compare performance of the models using the various inputs to show that the method using persistent homology is a strong option for investors on their stock market predictions and analysis.

Abstract Format

html

Language

English

Format

Electronic

Keywords

Stock exchanges—Mathematical models; Homology theory

Upload Full Text

wf_yes

Embargo Period

4-14-2028

Download

Available for download on Friday, April 14, 2028

Share

COinS