RULE-BASE EXPERT SYSTEM FOR EARTHQUAKE PREDICTION

Abstract

Data mining discipline is relatively newer as compared to conventional electrical and computer sciences, primarily because its evolution is outcome of higher processing powers and large amount of electronic data records. Together they reveal the new incredible truths of reality apparently hidden from our sights. The major focus areas of this subject includes: social web, e-commerce, retail/ item business, medical science, fraud detection and crime tracking. However, relatively less work is done against nonmedical scientific data. These enriched ‘scientific data mines’ remain almost unexplored so far. Seismic data of our planet falls in the same category. About two hundred distinct quakes of various magnitudes are recorded on this planet every day. According to the US Geological Survey from 2000 to 2010 there were average 63,000 deaths per year due to earthquakes globally. Earthquakes are among most terrifying of natural phenomenon that needs emergent attention. Striking without warning, and apparently coming out of nowhere; challenging our inherent assumption about stability of our planet. A mild quake may not be of much interest but a strong quake do carries awesome destructions. Aftereffects of seismic activity can resonate in human affairs for years, decades and centuries. The subject is quite challenging and researchers have been analyzing it from any potentially perceivable angle, like: mathematical modeling, data mining, hydrology analysis, ionosphere analysis and even animal responses. All current work on earthquake is restricted to a particular region only and don’t handle entire planet as one entity. Most of present prediction work is focused on predicting different types of potential damage assessments after a quake has actually occurred, where research part is more biased towards Geographical Information System (GIS) and Artificial Intelligence (AI). Some research work do focuses on prediction of aftershocks, being more predictable, as time series data analysis. This research focuses on the mining and analysis of attributes recorded for an earthquake by National Earthquake Information Center (NEIC) a subsidiary of United States Geological Survey (USGS). Investigation is carried out with objective to help and improve understanding of the phenomenon and support predictions in a way that improves the way we handle this disaster. This work, however, looks on quake attributes only to identify hidden relationships between them, using methods not known to be tried for predicting earthquakes and that are helpful in understanding a prospective quake and its behavior patterns. Results from our analysis clearly depict the effectiveness of our proposal.