| Summary: | 碩士 === 銘傳大學 === 資訊管理學系碩士班 === 93 === The former researches are always limited their analysis on the financial reports or focus on partial economical factors in considering the prediction factors of the return for the stock price. It can’t be applied to the real practices for limited factors. This research integrates the financial ratios and related economical factors such as interest rate, exchange rate and changes of the prosperity to predict the changes of stock price through the data mining techniques.
. We find out the most important variables from the thirty-seven variables by using the software of the Clementine. Then finds the variables that can promote the degree of the prediction accuracy. Besides, in the part of the clustering, it is the way that clusters all data into six groups by the characters of the data and then it uses the neural network to establish the model.
The highest accuracy degree of the prediction is 80.42 percent and the average of subsidiary prediction rate are also over 70 percent, that is better than former researches that only use financial ratios. In this research we found that the macroeconomics factors are the significant factors of prediction model.
|
|---|
