Multi-task clustering for stock selection to enhance prediction performance over multi data
Cited 0 times inCited 0 times in
- Multi-task clustering for stock selection to enhance prediction performance over multi data
- Bang, Eun Ji
- Kim, Kwang In
- Issue Date
- Graduate School of UNIST
- Stock prices are treated as non-stationary and time-variant data, because of generated by interest of various market participants. Also, research on stock prediction methods has conducted only their own individual or whole historical stock data, ignoring the fact that each price of financial instruments has a mutual organic relationship. Although recent research, propose adversarial learning methods to improve the generalization of the predictive model, but how to employ the correlated impacts of multiple dataset still remains an open problem. To solve this problem, we explore how to improve stock prediction performance by exploring multiple data. We introduce multi-task clustering method to incorporate highly correlated individual stock price data. The proposed method has extensively experimented with actual financial instrument data. Our novel methods outperform the previous state-of-the-art method with an average 1.66% improvement with respect to accuracy.
- Department of Computer Science and Engineering
- Go to Link;
- Appears in Collections:
- Files in This Item:
Multi-task clustering for stock selection to enhance prediction performance over multi data.pdf
can give you direct access to the published full text of this article. (UNISTARs only)
Show full item record
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.