Abstract: In view of the functions of monitoring system of tailing dam，combined with the content and features of safety monitoring data on tailing dam，and based on full understanding on the data mining techniques and its application，a forecasting method of infiltration route based on univariate time series of support vector regression is proposed.That is to select valid samples from the monitoring data of the tailings dam，and optimize the kernel parameters for SVR (Support Vector Regression) by using the leave-one-out method to establish a forecasting model.The results show that a higher accurate forecasting on the infiltration route can be obtained with few samples by this method.

Key words: Tailing dam, Safety monitoring, Data mining, Support vector regression