Social media sentiment enhanced stock market prediction analysis

• Main Author: Tan, Adrian Yong Chang
• Other Authors: Li Fang
• Format: Final Year Project
• Language: English
• Published: 2018
• Subjects: DRNTU::Engineering::Computer science and engineering
• Online Access: http://hdl.handle.net/10356/74014
• Institution: Nanyang Technological University

Stock market prediction is one of the key research areas in machine learning, providing researchers with a challenging task to forecast accurately, given that the dynamic nature of stock market and its relationship follows non-linearly. Various theories have highlighted the unattractiveness of stock market prediction, making it a redundant task. Therefore, this project aims to reflect the possibilities of stock market prediction in achieving high performance. We aim to explore the role of the learning-based methods in stock market prediction, and the reasons behind achieving high performance. For the implementation of the project, we randomly chose 5 stocks from Standard & Poor (S&P) 500 index via Yahoo Finance and used different learning-based methods to compare performance for both classification and regression problems. In addition, we split the dataset into training set, which comprises 70% of the dataset to build the model, and testing set, which comprises the remaining 30% to verify its performance. The result implies that for classification, ensemble learning methods tend to perform better in terms of accuracy, while SVM tend to perform better in terms of F-Measure. As for regression, all the learning-based methods used achieve high accuracy of more than 97%. In addition, based on the results, regression has a higher suitability in stock market prediction as compared to classification. In conclusion, this project highlights the possibilities of using machine learning for stock market prediction in achieving high performance. Therefore, it can contradict various theories that highlights the unattractiveness of stock market prediction, and more research efforts can be implemented on stock market prediction.