A data analysis algorithm based on statistical filtration and linear discriminant analysis

The rough sets reduction algorithm is an NP problem and is sensitivity to data noise. In order to overcome these disadvantages and deal with stochastic data, attributes selection and reduction is conducted by statistical filtration in this paper. This step corresponds to the process of attaining reduct of information system or decision system in the rough sets theory. Then the linear discriminant analysis classification algorithm based on Mahalanobis distance is used for classification, which is suitable for classification problem where its eigenvalue possess different type of property. Several databases of UCI repository are tested, using crossover validation. The results show that compared with C4.5 and SORES, the algorithm needs less property and produces less error recognition result. Furthermore the algorithm is simple and doesn´t need to quantify the continuous properties, which prove the validity of the algorithm