<![CDATA[Climate change that happens because of global warming also cause change in rainfall patterns. Knowing rainfall patterns is really important for some activity and works. So, rainfall forecasting is needed to understand the rainfall patterns in the future. One of the method used in forecasting is Support Vector Regression. But, SVR still has weakness in determining the right values for the parameters. So, an optimization algortithm is needed to help determining the values of the parameters in SVR. The purpose of this research is to do rainfall forecasting in Pujon area, Malang using Support Vector Regression that's been optimized by Improved-Particle Swarm Optimization. Optimization of SVR is done for getting the optimal values of SVR's parameters. The optimized SVR's parameters are (learning rate constants), (complexity), (Hessian's coefficient), (error rate) dan (kernel's coefficient). The rainfall forecasting for the first ten days of January from 2007 until 2015 by using IPSO-SVR resulted value of 0.213389 in RMSE compared to using only SVR which resulted value of 25.839085 in RMSE. This proved that optimization of SVR using IPSO is better compared to using the unoptimized SVR.]]>
Copyrights © 2017
