Abstrait

Ant Possibilistic Fuzzy Clustered Forecasting on High Dimensional Data

M.Ravichandran, A.Shanmugam

Stock market plays a significant role and has greater influence on basic economic energies of a country. Rapid changes in the stock exchange market with high dimensional uncertain data make the investors to look for effective forecasting using prediction mining techniques. The high dimensional stock data are classified into profitability, stability, cash flow and growth rate but does not deal completely with uncertain attribute values. On the other hand with large amount of uncertainty, the stock attributes and classes are not included simultaneously with the conditional probabilistic (i.e., Fuzzy set) distributional functions. Moreover, the test Possibilistic approaches (i.e., predictive mining) is not carried out on genuine uncertain data. So, the research pay attention on solving the forecasting problem with predictive data mining approach and helps the investors to select suitable portfolios. To forecast complex high dimensional uncertain data, Ant Possibilistic Fuzzy Clustered Forecasting (AP-FCF) method is proposed in this paper. AP-FCF method avoids the repeating mistake on uncertain stock attributes and classes and provides domain knowledge to the investors according to the current feature salience.