The Correlation Mining and Prediction of Social Security Events Based on Multidimensional Time Series Model

In recent years the frequent occurring of social security events has leaded a serious damage to the masses´ life and property. Based on large scale online time series data, this article mines event trigger factors and quantitatively analyzes their correlation to social security events by using multidimensional time series model. In addition, a situation dominated similarity measure method is presented to calculate the degree of similarity to events´ development trend. The experiments of analyzing 3 specific kinds of social security events show that the invisible trigger factors can be well mined and accurately predict the number of events may happen in the future. This can provide a new thought and method for the administrator to control and prevent these events from happening.