A New Heuristic Reduct Algorithm Base on Rough Sets Theory
Jing Zhang,Jianmin Wang,Deyi Li,Huacan He,Jia-Guang Sun
DOI: https://doi.org/10.1007/978-3-540-45160-0_24
2003-01-01
Abstract:Real world data sets usually have many features, which increases the complexity of data mining task. Feature selection, as a preprocessing step to the data mining, has been shown very effective in reducing dimensionality, removing irrelevant data, increasing learning accuracy, and improving comprehensibility. To find the optimal feature subsets is the aim of feature selection. Rough sets theory provides a mathematical approach to find optimal feature subset, but this approach is time consuming. In this paper, we propose a novel heuristic algorithm based on rough sets theory to find out the feature subset. This algorithm employs appearing frequency of attribute as heuristic information. Experiment results show in most times our algorithm can find out optimal feature subset quickly and efficiently.