Publication

Abstract : Studies show that in past few years out of four deaths, one is due to cancer. The area of molecular oncology witnessed the involvement of small non-coding RNAs known as MiRNAs in these complex disorders. Using MiRNA expression profiling, the stages of complex diseases can be analyzed, and suitable treatment strategy can be devised from the very beginning. The current study focuses on the importance of performing feature selection before classification. Genetic Algorithm (GA) is used for feature selection to select a set of MiRNAs from a large dataset that can classify early and advanced stages of breast cancer. Two classifiers, namely Support Vector Machines (SVMs) and random forest, were analyzed here. The result signifies that random forest classifier performs much better than SVM, and feature selection improves the prediction accuracy. SVM performance was studied by using different kernel functions and concluded that polynomial kernel gives a better result.