Accurate Prediction of Significant Earthquakes using Machine Learning Algorithms

Hawalli Ramachandra, Samrudhi (2024) Accurate Prediction of Significant Earthquakes using Machine Learning Algorithms. Masters thesis, Dublin, National College of Ireland.

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

Abstract

Earthquakes are, probably, the most damaging of all the natural disasters. Due to the nature of seismic activities, it’s tough to predict them more or less accurately because of the various inherent limitations that are present within the traditional statistical methods. This paper aims to explore the probability of machine learning algorithms in helping the accuracy of earthquake prediction by using the attributes of geospatial and temporal data. To meet this requirement, three datasets sourced from Kaggle have been used: ”Earthquakes 2023 Global”, ”Earthquake Dataset," and ”Global Earthquake and Aftershock Data (January 2023)”. All three have rich data characters such as magnitude and depth as well as spatial coordinates along with temporal data that aids for more complex models. The machine learning models tested include ensemble methods like Random Forest and Gradient Boosting; linear models, including Linear Regression and Lasso Regression; and even more complex models, like Support Vector Regression. Among all the models, Random Forest always has been the best performer where R² values have been greater than 0.90 and low Mean Squared Error value. Geospatial and time-related features such as latitude, longitude, and recurrence interval dominated the model’s performance. Further insight into the interplay between the features used and the generated seismic patterns was captured through visualizations including heatmaps and scatter plots. This study reveals unprecedented machine learning-based potential in predicting earthquakes while underlining the need to combine spatial and temporal aspects for better performance. Therefore, the results call for real-time embedding of predictive models into seismic monitoring frames for better preparedness of disasters. More work in advanced techniques of deep learning, richer data sets, and multi-hazard frameworks will further develop earthquake prediction methodologies.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Menghwar, Teerath Kumar UNSPECIFIED
Subjects:	G Geography. Anthropology. Recreation > GE Environmental Sciences Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science G Geography. Anthropology. Recreation > GE Environmental Sciences > Earth sciences > Geology > Physical geology > Sedimentation and deposition > Earth movements > Earthquakes 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 12:44
Last Modified:	04 Sep 2025 15:41
URI:	https://norma.ncirl.ie/id/eprint/8709

Actions (login required)

View Item