Abstract: Based on the wavelet analysis of earthquake data in Mainland China for the recent century, we determined the characteristics of earthquake activity in different time scales. Meanwhile, the main periods in earthquake activity are obtained from different wavelet coefficients, and as a parameter, the periods are used for earthquake forecasting by means of least square support vector machine (LS-SVM). The results show that this method has a high success rate with smaller average error.

Key words: Earthquake forecast, LS-SVM, Wavelet, Non-linear time series