<![CDATA[Stock is a proof of investing in a corporation and stock holders have the right to claim part of corporation's earning and assets. Stock holders can gain a lot ot of benefit such receiving dividens and selling their stocks with higher value (capital gain). Stock holders need to be careful to manage their assets because stock prices keep changing over time. Stock holders usually monitor stock prices change and analyze them by forecasting. Support Vector Regression (SVR) is one of forecasting methods that performs well in both linear and non linear data. SVR can obtained a fitted model that is neither overfit nor underfit. However SVR has one drawback. The performance of SVR is greatly affected by its parameter. So finding the right parameter value on SVR is needed to gain a good forecasting result. One of optimization algorithms is Genetic Algorithm. Genetic Algorithm is used in order to get the right value of SVR parameter. SVR that is optimized by Genetic Algorithm is capable of getting a good result in forecasting. The test shows error rate/MAPE of forecasting is 0.165% which is smaller than using SVR which is 1.612% with best parameters such as population size 50, generation 200, crossover rate 0.4, mutation rate 0.6, range of sigma 0.5-1, range of epsilon 10-7-10-3, range of C 0.001-5, and range of gamma 10-5-10-3.]]>
Copyrights © 2018
