Combining the mRMR technique with the Northern Goshawk Algorithm (NGHA) to choose genes for cancer classification

This paper presents a novel strategy for gene selection in cancer classification by integrating the Northern Goshawk Algorithm (NGHA) with microarray datasets. The primary objective is to enhance the precision of cancer diagnosis by efficiently identifying informative genes within the often-noisy nature of these datasets. The proposed method consists of two key stages: the filter stage, employing the Minimum Redundancy Maximum Relevancy (mRMR) method, and the wrapper stage, harnessing the synergies of NGHA and Support Vector Machine (SVM) classifier. Experimental assessments on two microarray datasets demonstrate the method's accuracy and effectiveness. Comparative evaluations against common gene selection techniques indicate comparable performance on several datasets, with particularly promising and novel results on two datasets. The innovative integration of mRMR and the Northern Goshawk Algorithm presents a potent strategy for advancing gene selection in cancer classification, holding significant potential for elevating diagnostic precision in oncology.