Enhancing Cryptocurrency Price Prediction using Transformer-Based Models for effective Time-Series Analysis

Karicheti, Rambabu (2024) Enhancing Cryptocurrency Price Prediction using Transformer-Based Models for effective Time-Series Analysis. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (1MB) | Preview

Abstract

Cryptocurrency prices highly fluctuating and accurate price prediction of cryptocurrency is difficult but important task for the investor or trader. The conventional machine learning models are not well equipped to capture the sequential and temporal relationships which are present in cryptocurrency time-series data. This work aims at analyzing the of using transformer-based models that are good at modeling long-term dependencies and intricate structures for predicting cryptocurrency prices. The emphasis is on the method which helps to forecast the Bitcoin prices based on the historical prices and other characteristics of the market. The forecasting performance of the transformer model is compared with other baseline models such as SVM, Gradient Boosting (GBM), Random Forest (RF), RNN and LSTM. Performance metrics such as MSE, MAE, RMSE, and R² are applied to compare the accurate predictions of the model. The transformer model outperformed all other models with an MSE of 123709.59, MAE of 295.12, RMSE of 351.72 and R² of 0.9801. In comparison, traditional models such as SVM, Random Forest, Gradient Boosting and deep learning models like RNN and LSTM are unable to recognize the long-term dependencies and patterns in the change in the price of cryptocurrencies. These results demonstrate that the transformer model outperforms other models in the forecast of highly unpredictable cryptocurrency prices and positions them as a potentially viable solution for enhancing the accuracy of financial predictions. It also highlights directions for future research, such as the including of the other features in the market, real-time prediction models, and more comprehensible models, to aid both theoretical and applied research in cryptocurrency price prediction.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Khan, Sallar UNSPECIFIED
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science H Social Sciences > HG Finance > Money > Digital currency > Cryptocurrencies Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	02 Sep 2025 15:56
Last Modified:	02 Sep 2025 15:56
URI:	https://norma.ncirl.ie/id/eprint/8724

Actions (login required)

View Item