| Summary: | 碩士 === 銘傳大學 === 應用統計與資料科學學系碩士班 === 107 === In recent years, with the development of our economy and information technology, more and more people have begun to use new technologies to obtain the resources of financial. There are also a large number of methods for investing in wealth management on the Internet, providing the public as a reference. To many research methods for stock market fluctuations which has its own characteristics, but most of them significant focus on fundamental and technical analysis. These two analysis methods ignored the impact of external information changes on stock prices. Based on the stock price of Taiwan's semiconductor industry, this study attempts to use the text mining to forecast the stock price on the financial news. Discussing the relation to Semiconductor stocks in Taiwan and all the variables by using the results of the macroeconomic variables and text mining which affecting the stock price as influencing factors.
There were 524 trading days from January 2016 to March 2018 during the study period.
We selected MTK, TSMC and ASE as the object of the research. We can make a prediction by sorting the data of stock price and financial news in the past years.
The empirical analysis can lead to the following conclusions: In the aspect of stock prediction, the highest predicted values among the three semiconductor stocks was 42.37%, is the reverse effect. In the complex regression analysis, if the overall economic factors are removed, the news will have a weak negative impact on the stock price. In the structural equation model, semiconductor stocks are relevant. In vector auto-regression model, MTK were the only one that affected by news.
|
|---|
