Skip to Main Content
Frequently Asked Questions
Submit an ETD
Global Search Box
Need Help?
Keyword Search
Participating Institutions
Advanced Search
School Logo
Files
File List
osu1321989426.pdf (714.44 KB)
ETD Abstract Container
Abstract Header
Detecting Self-Correlation of Nonlinear, Lognormal, Time-Series Data via DBSCAN Clustering Method, Using Stock Price Data as Example
Author Info
Huo, Shiyin
Permalink:
http://rave.ohiolink.edu/etdc/view?acc_num=osu1321989426
Abstract Details
Year and Degree
2011, Master of Science, Ohio State University, Computer Science and Engineering.
Abstract
In our modern world, the ability to predict the trend of nonlinear, irregularly shaped data is of great importance. This analysis can be used in many industries, such as in the prediction of stock prices, predicting a certain region’s future electrical usage to make the grid more efficient and so on. This study is focused primarily on predicting rising and falling after a smooth period when the time series data shows a “turning point”. The nonlinear, irregularly shaped time series data which we study in this thesis always show consistent and repeated rising and falling pattern over time. Thus we can predict the future rising and falling via studying the past data. This requires analysis of the self-correlation between the past time-series data. In the last twenty years, many methods have attempted to predict this very aspect. Although some of these methods have shown good predictive accuracy, the high computational cost and low efficiency made them impractical outside the academic scope. In this thesis, I have proposed an efficient method for analyzing self-correlation of a nonlinear, irregularly shaped time-series data using stock data as an example, which will select stocks and perform automatic trading. Rather than attempting to predict data future rising and falling at every moment, this method only runs the prediction function when some “turning point” becomes apparent. The DBSCAN clustering method is used to obtain self-correlation between current and previous segments of time series data. By calculating the average expected rate of return for similar segments, we derive the predicted parameters. To select the optimal stock for trading, we need to simulate how the stock will perform in the future with the predicted parameters. At last, we select the stock whose future performance is best. The contributions of this thesis are as follows: 1 This thesis proposes a method to automatically locate data peaks and valleys. 2 This thesis also proposes a method to analyze self-correlation of nonlinear, irregularly shaped time series data with highly reduced computational cost. 3 Finally, optimal stock selection is achieved by analyzing the stock’s predicted expected rate of return and its volatility.
Committee
Rajiv Ramnath (Advisor)
Jayashree Ramanathan (Committee Member)
Pages
93 p.
Subject Headings
Computer Engineering
Keywords
DBSCAN clustering
;
Monte Carlo simulation
;
Trading Strategy
Recommended Citations
Refworks
EndNote
RIS
Mendeley
Citations
Huo, S. (2011).
Detecting Self-Correlation of Nonlinear, Lognormal, Time-Series Data via DBSCAN Clustering Method, Using Stock Price Data as Example
[Master's thesis, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1321989426
APA Style (7th edition)
Huo, Shiyin.
Detecting Self-Correlation of Nonlinear, Lognormal, Time-Series Data via DBSCAN Clustering Method, Using Stock Price Data as Example.
2011. Ohio State University, Master's thesis.
OhioLINK Electronic Theses and Dissertations Center
, http://rave.ohiolink.edu/etdc/view?acc_num=osu1321989426.
MLA Style (8th edition)
Huo, Shiyin. "Detecting Self-Correlation of Nonlinear, Lognormal, Time-Series Data via DBSCAN Clustering Method, Using Stock Price Data as Example." Master's thesis, Ohio State University, 2011. http://rave.ohiolink.edu/etdc/view?acc_num=osu1321989426
Chicago Manual of Style (17th edition)
Abstract Footer
Document number:
osu1321989426
Download Count:
1,319
Copyright Info
© 2011, all rights reserved.
This open access ETD is published by The Ohio State University and OhioLINK.