<![CDATA[Financial time series is one of the most challenging applications of modern time series forecasting. Financial time series is closely related to noise (disturbance signals), non-stationary and deterministic chaotic. The characteristics indicate that no exhaustive information can be obtained from the past behavior of financial markets to capture the full dependency between future prices and that of the past. The purpose of this study is to determine the correct algorithm for Artificial Neural Networks, Multiple Regression Analysis and Support Vector Machines so that the highest accuracy value for prediction trend accuracy and the smallest value for Root Mean Square Error can be obtained. The research method used is 1. Neural network, 2. SVM, 3. Linear regression. From the data that has been tested using the Rapidminer tool with parameters that refer to several literatures [5], data collection is taken from the Stock Market Online Application "MetaTrader version 4" of the type "daily/Daily" with a time span of "03/09/2001 up to 25/07/2012", a total of 2052 data", with the attribute "Date, Open, High, Low, Close, Volume" with the Main attribute "Close" using the Support vector machine algorithm, artificial neural network and multiple linear regression, then the conclusion drawn is that the value close to the series value is the value tested on the support vector machine algorithm, with parameters For RMSE values close to the value "0" obtained from the measurement results in the SVM Algorithm on the RBF (radial basis function) kernel with a value "gamma" γ = 100 with RMSE = 0.000, and SE = 0.000. with prediction accuracy error = 0.976]]>
Copyrights © 2023
