Analysis of earthquakes by means of data mining methods

The goals of this study can be summarized as analysis of seismic activities through data mining methods and the development of software which can be used to predict earthquakes before they take place. In this regard four different implementations about the seismic data are realized.The first application is the Turkey Earthquake Portal (TDEP) which is created to carry out statitical processing, querying and data mining studies on seismic data. Prediction of earthuakes is aimed by interpreting the statistical information about the seismic data. Significant advantages of the application in terms of usage and interpretation is observed.The second application consists of the study which is developed to reveal the hidden associations between earthquake zones. In this application one of the data mining techniques, association rules are successfully applied on the seismic data and various rules are extracted. Evaluation of these rules shows that the system produces successful results.In the third application a novel approach to earthquake prediction is proposed. In paralel to this approach a prediction algorithm is designed and successfully applied to seismic data. Evaluation of the prediction results shows that the algorithm produces successful results.The fourth application consists of the clustering of seismic data. In this section clustering results of earthquakes occured in Turkey is discussed. For clustering k-Means and K-NN algorithms are effectively used. An alternative approach to traditional k-Means initial cluster center selection is proposed in this section for seismic data. It is observed that this approach demonstrates better performance in data clustering than the traditional k-Means algorithm.