Damage to infrastructure, human casualties,

and the economy can all take a nosedive when an earthquake

strikes, making it one of the deadliest and most unpredictable

natural disasters. Factors such as height, ground motion,

rainfall, rock bed material, and area tectonics all have a role

in their occurrence. Geologists and seismologists still have a

significant obstacle when trying to predict when, where, and

how big of an earthquake will be. Using data preprocessing

techniques like transformation and cleaning in conjunction

with advanced feature extraction methods based on

precursory patterns, this study presents an Earthquake

Prediction System. The system expands upon previous

research that found precursory indicators like changes in

animal behaviour, increasing temperatures, radon gas

emissions, and variations in seismicity. Particular attention

was given to the LSSVM-FPA hybrid model, which integrates

the LS-SVM and the FPA, in the development of two

predictive models that make use of seismic indicators and

hybrid ML techniques. When compared to other models, the

LSSVM-FPA model outperforms them all with a prediction

accuracy of 97.43%. In light of these findings, it is reasonable

to conclude that the proposed Earthquake Prediction System

provides a solid basis for enhancing early detection and

reducing the devastating effects of future seismic events