Data mining classifiers comparison for seismic hazard prediction

Abstract

Earthquake and seismic hazards are natural disasters which are very difficult to predict. Researchers are working hard to predict these disasters for minimizing loss of life and property. Proposed research used data mining algorithms on seismic bumps dataset which was obtained from coal mines for the seismic hazard prediction. Data mining is a powerful technique used to discover patterns of data. In this research, performance of five data mining classifiers was compared for better prediction of seismic hazard. For preprocessing of this dataset, discretization and resampling techniques were used. For modelling, five data mining classifiers were implemented and compared by using feature selection technique on the basis of confusion matrix measures like success rate, mean absolute error, kappa statistics, precision, recall and f-measure. This analysis showed that Random Forest algorithm achieved highest success rate by using feature selection technique and provided promising results for seismic hazard prediction.