Building Decision Rules Using a Novel Data Driven Method for Microarray Data Classification

Knowledge gained through classification of microarray gene expression data is increasingly important as they are useful for phenotype classification of diseases. Different from black box methods, rule based system can produce interpretable classifier with knowledge compressed in terms of rules. This paper proposes a rule based approach called "Large coverage rule (LCR)" for microarray data classification. The proposed approach is a parameter free data driven approach that constructs decision rule based on the expression values of a gene. A simple "Rank-Based Scoring (RBS) algorithm is proposed for selecting informative genes. The performance of the proposed approach is evaluated using four gene expression data set. From the simulation result, it is found that the proposed approach generates compact rules and produces comparatively good classification accuracy than the other approaches reported in the literature. Stability analysis of the test result shows that the rules generated by the proposed LCR approach are simple to interpret and are highly comprehensible.