Processing and analyzing data to predict earthquakes in Iraq

Abstract

The An earthquake is a devastating natural disaster that causes great economic and human losses because it occurs without warning. The increase in earthquakes in Iraq has raised concerns about the future of the region. It is necessary to study earthquake prediction and determine the location, size and time of the earthquake. A machine learning model was proposed to predict earthquakes in Iraq using two sources: the first is a catalog of data from 1900 to 2019, which includes 36,663 earthquakes, and the second is from the USGS for one year from 2022 to 2023, which includes 25,000 earthquakes. Preliminary processing of the data was done, removing outliers and integrating Date and time data in timestamps, and the five important features for prediction were identified and the data was divided into 80% for training and 20% for testing. After applying several attempts in using different models, the best results were achieved using NN and the accuracy was about 0.7. The most important reason for this result is training over many years that may change geologically. The study compared its results with other studies to predict earthquakes across different regions of the world.