The Study of Elman Recurrent Neural Network in TSEC Taiwan 50 Index

• Main Authors: Tsai, Ping Yang, 蔡炳煬
• Other Authors: Shiue, Yeou-Ren
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/29097879021137902554

碩士 === 華梵大學 === 資訊管理學系碩士班 === 99 === In this study, the Elman neural network feedback artificial neural networks(Elman Recurrent Neural Network) for the model to the Taiwan Securities Exchange was the subject of Taiwan 50 Index, the closing index of the next day forecast, and with the back-propagation neural Network(Back-Propagation Neural Network, BPN) as a model to do an analysis to explore. Scope of information to the Taiwan stock exchange since January 1, 2003 until September 30, 2010 only. Taiwan 50 Index data, as well as international stock markets, Taiwan stocks after-hours information combined with technical indicators, as neural network input variables. Elman and by the analysis of BPN, in days, for testing, training, stock market forecasts of Elman network as is feasible. The results show that the overall performance of the Elman network is better than BPN; Elman of the MSE of 1.17E-08 better than the BPN of 1.04E-03, so; Elman neural network for forecasting the Taiwan 50 Index closed the next day is feasible, effective .