| Summary: | 碩士 === 國立成功大學 === 統計學系碩博士班 === 92 === This paper is to study the application of artificial neural network and statistical method in forecasting tendency of stock market price index and analysis of inner characteristic of stock price trend by using major influence factor of stock price index. According to statistical data that is derived from Taiwan Stock Exchange Corporation, the electronics industry is mainstream industry. For this reason, electronics industry Stock price index is our research target. However, in past research and paper, little is consider about the stock price index by complete variable and approach. Thus, this paper tried to contain effect factor of stock price, including technical variable、macro-economical variable and industry basic variable.
This paper using above variable establish prediction model by regression analysis、time series and back-propagation network forecast tendency of stock market price index in the future. By the same using above variable classify stock price by cluster analysis, and then using discriminate Analysis and probabilistic neural network determine error rate in all groups. Finally, comparing accuracy of all the prediction and classification models.
Experimental result reveal in accuracy of the prediction and classification model. Respectively, the back-propagation network models are best, the second are regression models and the worst are time series models; discriminate Analysis and probabilistic neural network have the same result.
|
|---|
