|
|
 |
| |
|
|
|
Machine Learning-Based Earthquake Prediction: Feature Engineering and Model Performance Using Synthetic Seismic Data |
|
|
|
PP: 695-702 |
|
|
|
doi:10.18576/amis/190317
|
|
|
|
Author(s) |
|
|
|
Amena Mahmoud,
Othman Alrusaini,
Emad Shafie,
Ayman Aboalndr,
Manal S. Elbelkasy,
|
|
|
|
Abstract |
|
|
| Earthquake prediction is still a nightmare in terms of minimizing losses during seismic catastrophes. This research seeks to apply machine learning techniques for classifying and forecasting earthquakes using synthetic seismic records. It is observed that key time-series features including Root Mean Square (RMS) amplitude and spectral peak frequency were extracted from time-series waveforms and then used for training a Support Vector Machine (SVM) classifier. The model was able to attain an accuracy rate of 90% which shows how efficient the presented features were in distinguishing different seismic events. There was a geographic visualization of predicted events that generated insights that were useful in locating places prone to seismic hazards. The synthetic database engendered a laboratory-like test setting, but the shortcomings in practical relevance underline the necessity for inclusion of real earthquake data sets. This research adds to the increasing body of literature about data driven seismic analysis and paves the way to strengthen predictive models which would contribute to better earthquake preparedness. |
|
|
|
|
|
|
|
