Opinion based intelligent stock market prediction

• Main Author: Shi, Yilin
• Other Authors: Lin Zhiping
• Format: Final Year Project
• Language: English
• Published: 2019
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Computer hardware, software and systems
• Online Access: http://hdl.handle.net/10356/77276
• Institution: Nanyang Technological University

The stock market plays a significant role in financial markets. However, stock prices are affected by multiple volatile factors which result in highly complex time series that are in turn extremely challenging to correctly predict. Recently, scholars and industry researchers have invested much effort into exposing the principles behind the stock market price fluctuations through big data analysis; specifically, via machine learning. Some of the existing methods relying on machine learning methods have obtained fairly high accuracy in predicting certain stock market prices. Beyond stock prices over time, Public sentiment is one of the most important factors that can be factored into machine learning models to enhance the prediction outcome. According to research, public sentiment indeed affects the changes in stock market since it reflects investors’ attitude towards the companies. This project proposes to introduce a machine learning based method that can discover certain patterns in the price changes of stock markets and analyze the relationship between price changes and social sentiment. The proposed hybrid learning-based model is a combination of Long Short Term Memory model and Support Vector Regression model. The trend of the stock price is predicted, and historical stock data, sentiment data together with some economic indexes are included as input parameters. The report illustrates the full procedure of data collection, data processing and the model execution. Furthermore, the report demonstrates the effectiveness and performance of the proposed method, measured by accuracy. The report introduces the advantages of the proposed method and further discusses the future direction in this domain.