Earthquake forecasting with time series analysis

With the developing technologies, increase in the amount of produced data, very large data has begun to form. Data mining methods have been developed with the aim of obtaining useful information from these data. Data mining provides to obtain valuable information by applying different methods. Some of these methods are clustering, classification, association rule, time series etc. In this thesis, time series analysis method was mentioned and studies on earthquake prediction were made. Time series analysis provides predictions about future by working on past data. In the thesis, studies on earthquake prediction were made by using time series analysis and artificial neural network. Artificial neural networks are one of the commonly used methods in the analysis of nonlinear series. The use of artificial neural network for nonlinear time series is given in detail in the thesis. Earthquake data of Elazig province were used for earthquake prediction. These data were obtained from Boğaziçi University, Kandilli Observatory and analyzes were carried out by creating various data sets from these data. As a result of the analyzes, the success rate of the artificial neural network method was found to be high in the prediction of low-intensity earthquakes. However, the prediction rate of high-intensity earthquakes remained at a low level. As a result of the analyzes carried out, it was concluded that the artificial neural network is a method with high performance in nonlinear series. In addition, it has been seen that the lack of data and irregularity in the time series are the factors that prevent the desired success in earthquake prediction. Keywords: Data Mining, Time Series Analysis, Artificial Neural Network, Earthquake Prediction