Gene Feature Selection Method Based on ReliefF and Pearson Correlation

Based on the microarray gene expression data with high dimensional and small samples, this paper proposes a gene feature selection method based on ReliefF and Pearson correlation algorithm. Firstly, this method evaluates the feature weight by ReliefF, sorts the features according to the weight, and eliminates the irrelevant genes according to a given threshold. After that, use the Pearson correlation algorithm to eliminate the redundancy of the selected feature subsets for further feature selection. Finally, for the sake of verifying the feasibility and effectiveness of the above method, apply this method to three gene expression data sets, including Colon, Leukemia and Prostate, for feature selection. And two classifiers including k-nearest neighbor (KNN) and support vector machine (SVM) are used to classify and identify the sample data after dimension reduction. It is shown in the experiment that the method proposed in this paper can effectively eliminate unrelated and redundant genes, and achieve higher classification accuracy.