Hybrid Neural Network and Decision Tree for for Exchange Rates Forecasting

• Main Authors: Ardiansyah, Soleh, Mazlina, Abdul Majid, Jasni, Mohamad Zain
• Format: Conference or Workshop Item
• Language: English
• Published: 2012
• Subjects: QA76 Computer software
• Online Access: http://umpir.ump.edu.my/id/eprint/3518/1/6-ICoCSIM.pdf; http://umpir.ump.edu.my/id/eprint/3518/

As the largest financial market in the world, foreign exchange (Forex) is becoming a very profitable market with a daily transaction of more than 3.0 trillion U.S. dollars. Therefore, predicting about it has been a challenge for many years. Artificial Neural Network (ANN) provides better performance of forecasting but it tends to get stuck in local minima and there is no optimal way to determine the best classifier on it. Meanwhile, Decision Tree (DT) is able to generate classifier in the form of a tree. This paper proposes a hybrid prediction model by combining both ANN and DTalgorithm to predict exchange rates. The models are constructed by using the better of parameters and architectures based on related work such as filtering mechanism, number of hidden layers, number of hidden neurons, training algorithm, and error measurement, with the assumption that if the hybrid model is constructed by the better parameters and architectures, then the output of the model also produces better result