Hybrid feature selection model based on relief-based algorithms and regulizer algorithms for cancer classification
Cancer is a group of diseases that involve abnormal cell growth with the potential to spread to other parts of the body. Cancer microarray data usually include a small number of samples with a large number of gene expression levels as features. Gene expression or microarray is a technology that monitors the expression of the large number of genes in parallel that make it useful in cancer classification, high dimensionality in cancer microarray data results in the overfitting problem. This article proposes novel hybrid feature selection model called the RBARegulizer model, which is based on two types of feature selection techniques, two RBAs algorithms (ReliefF, MultiSURF) for feature-ranking filters to the most important one's genes, and three regulizer algorithms (Lasso, Elastic Net, Elastic Net CV) to reduce the feature subset, remove the noisy and irrelevant feature to improve the performance and accuracy of cancer (microarray) data classification. For evaluating the model, the different three classifiers SVM, MLP, and random forest with four high-dimensional microarray data for different cancer types were applied. The experimental type shows that our model overcomes the overfitting problem of cancer microarray data. Moreover, the results show that RBARegulizer model is perfect in improving the accuracy of cancer microarray data classification. © 2021 John Wiley & Sons, Ltd.