Primary component analysis method and reduction of seismicity parameters

Introduction Data mining (SHAO and YU, 2003) is a new kind of technique developed with database and artificial intelligence in recent years, which processes the data in the database to abstract the im- plied and pre-unknown, but potentially useful information and knowledge from large amounts of incomplete, noisy, blurring and stochastic data. For data mining, data purging is an important link beforehand that includes eliminating noise, making up lost domain, and deleting ineffective data, as…

Primary component analysis method and reduction of seismicity parameters

In the paper, the primary component analysis is made using 8 seismicity parameters of earthquake frequency N (M l≥3.0), b-value, η-value, A(b)-value, Mf-value, Ac-value, C-value and D-value that reflect the characteristics of magnitude, time and space distribution of seismicity from different respects. By using the primary component analysis method, the synthesis parameter W reflecting the anomalous features of earthquake magnitude, time and space distribution can be gained. Generally, there is some relativity among the 8 parameters, but their variations are different in different periods. The earthquake prediction based on these parameters is not very well. However, the synthesis parameter W showed obvious anomalies before 13 earthquakes (M _S≥5.8) occurred in North China, which indicates that the synthesis parameter W can reflect the anomalous characteristics of magnitude, time and space distribution of seismicity better. Other problems related to the conclusions drawn by the primary component analysis method are also discussed. Foundation item: Project of Joint Seismological Science Foundation of China (104090).