| Summary: | 碩士 === 國立中正大學 === 電機工程研究所 === 103 === A stock market generates a large amount of data at every moment during trading. If we can analyze the trading data rapidly and efficiently, then it can provide investors the prediction information. With the rise of computers, many experts and scholars try to analyze stock markets by artificial intelligence and then predict the price. Using computers in analysis has the advantage in real-time calculation of trading data. In this situation, finding a good way to analyze rapidly can make the investor more outstanding performance than traditional manual analysis. Besides, computers are suitable with a large among of calculation because of the characteristic of technical analysis. The indexes parameter selection is a very important issue in technical analysis and genetic algorithms are a good tool to choose parameters for it. It is a method which simulates biological evolution to find the valid index parameters from the system. These parameters can be used in the system effectively. However, a genetic algorithm spends more time than other artificial intelligence methods relatively. In order to solve this problem, it can adopt distributed computing to enhance its computational efficiency. Besides, a distributed genetic algorithm will create multiple independent evolution environment which can be dedicated to a specific gene. It can not only enhance the speed of calculation but also make the superiority of specific variable to be reserved. Therefore, this thesis will adopt a distributed genetic algorithm to do stock market analysis. We expect it can enhance the speed of genetic algorithm and still remain a good performance. Finally, we will evaluate our performance by calculation time.
|
|---|
