Enhancing Stock Market Predictions with Deep Neural Networks and Time Series Analysis

Rakesh, Pidugu (2024) Enhancing Stock Market Predictions with Deep Neural Networks and Time Series Analysis. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (2MB) | Preview
Preview	PDF (Configuration Manual) Download (903kB) | Preview

Abstract

The volatile and dynamic nature of the stock market presents both challenges and opportunities for investors and financial analysts alike. In recent years, the advent of advanced computational techniques and the availability of vast amounts of financial data have paved the way for the application of machine learning and deep learning algorithms in stock market prediction. This research study endeavours to explore the intersection of deep neural networks and time series analysis in the realm of stock market forecasting, with a focus on enhancing prediction accuracy and market understanding. The study begins by delving into the intricacies of historical stock market data, leveraging a publicly available dataset containing information on the price and trading activity of Microsoft Corporation (MSFT). Using the Comprehensive way of data preprocessing and exploratory data analysis, it becomes possible to detect the main valuable patterns, trends, as well as the relationships between various factors that influence the stock market’s overall behavior. Thereafter, the research progresses to the development and evaluation of the forecasting model which uses LSTM, GRU, ResNet, and RNN architectures in the deep learning with the help of data fusion approach. These models are induced to learn from past stock prices which are forecasted to pinpoint specific future movements incorporating time dependencies and nonlinear relationships inherent in financial time series data. In addition, this research also aims to test the efficiency of hybrid model that are based on various architectures used earlier, which have already been trained, but now again we take all the same independent deep learning models of LSTM, GRU, ResNet and RNN architecture models to form hybrid model by combining these architectures achieved the outstanding competitive results with an RMSE of 233.35 and a MAE of 229.18 then the particular independent Model. Mechanisms include exploitation and differentiation of their respective advantages in order to strengthen the accuracy of forecasts. Through the use of theories and techniques from different models and methods, providing strategic recommendations and insights to stakeholders within finance sector could prove to be a better decision-making strategy as well as risk management in finance sector. In summarization, this study research provides additional data to the accumulative knowledge bank on stock market prediction through showing the efficiency of deep neural networks and time series analysis methods in enhancing accuracy prediction. The study shows that deep learning algorithms can be employed to discover such valuable insights and provide worth investment strategies to the finance world which is now characterized by more complexity and unpredictability.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Shahid, Abdul UNSPECIFIED
Uncontrolled Keywords:	Stock Market; Microsoft; Deep Learning models; LSTM; RNN; GRU
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning H Social Sciences > HG Finance > Investment > Stock Exchange
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	05 Jun 2025 15:21
Last Modified:	05 Jun 2025 15:21
URI:	https://norma.ncirl.ie/id/eprint/7770

Actions (login required)

View Item