Improved Hybrid Model Based on Support Vector Regression Machine for Monthly Precipitation Forecasting

In this paper, we study the time series techniques for the monthly precipitation forecasting. The techniques used in this study are moving average procedure, support vector regression machine, and seasonal autoregressive integrated moving average model and hybrid procedure. Firstly, the moving average procedure is employed to find the trend; secondly, the support vector regression machine is applied to forecast the trend; thirdly, the hybrid procedure is used for provide the last forecasting results based on the above models. For the coefficients, the optimization method we employed is the popular particle swarm optimization algorithm. Three time series are applied to test the proposed idea, which are the monthly precipitation data from Gansu Meteorological Bureau. The forecasting results show that our proposed model is an effective model for nonlinear time series forecasting.