A methodology for discovering relationships from stock and economic data utilizing rough set theory is presented. The components of the methodology are the scope definition, data collection, discretization, rule generation, and verification. Two applications integrating rough set theory for stock market analysis are reviewed. The applications are examined according to the methodology defined in this thesis. These applications discover relationships in stock market data utilizing a knowledge discovery tool. Monthly stock and economic data over the 1980s is analyzed. The first application identifies dominant relationships among fluctuations of the monthly stock market indicators. The second application discovers temporal relationships going back in time for 6 months. Verification is established through a domain expert, a stock broker, who confirms the validity of the relationships discovered. This thesis defines a methodology which establishes rough set theory as a good candidate for knowledge discovery with stock market data.
展开▼
