Using Parallel Genetic Algorithm to Create a Stock Portfolio on Hadoop Platform

• Main Authors: Li-Bang Chen, 陳立邦
• Other Authors: Kuo-Chen Li
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/md7799

碩士 === 中原大學 === 資訊管理研究所 === 105 === Due to economic depression and the expansion of Asia Pacific Finance, more and more people start to learn the concept of investment about financial management. On account of financial market and the diversity of investment tools, the investors are facing more complicate investment environment. It is very difficult to choose the effective investment targets. In addition, with the development of internet, big data is accumulated and requires novel algorithms to analyze the data. Therefore, the distributed frameworks such as “Hadoop” platform are widely used to approach these problems. This research focuses on the use of technical indicators on Taiwan stock market based on big date analysis. The system provides investors strategies by using distributed genetic algorithm on technical indicators. The recommended investment combination could help investors to distribute the risk of investment. The research is built on the distributed Hadoop framework to be able to process the huge transaction data and combine with technical indicators with parallel genetic algorithm. The findings also established an investment combination selecting model that supports investors without domain knowledge to make decisions. Through this model, the investors could input the preferences for the individual investment approaches, and system will calculate and select the recommended investment combinations as decision supports. To validate the proposed model, the Preliminary experiments are carried out and the results show the effectiveness of the system.