Attribute selection based on information gain ratio in fuzzy rough set theory with application to tumor classification

Tumor classification based on gene expression levels is important for tumor diagnosis. Since tumor data in gene expression contain thousands of attributes, attribute selection for tumor data in gene expression becomes a key point for tumor classification. Inspired by the concept of gain ratio in decision tree theory, an attribute selection method based on fuzzy gain ratio under the framework of fuzzy rough set theory is proposed. The approach is compared to several other approaches on three real world tumor data sets in gene expression. Results show that the proposed method is effective. This work may supply an optional strategy for dealing with tumor data in gene expression or other applications.

Figure optionsDownload as PowerPoint slideHighlights
► We introduce gain ratio into fuzzy rough set model.
► We propose an attribute reduction method based on information gain ratio under the framework of fuzzy rough set theory.
► The proposed approach is used to reduce attributes in tumor data in expression of gene. Experiments results on real tumor data verify the effectiveness of the proposed approach.