Hybrid Filter Feature Selection for Improving Cancer Classification in High-Dimensional Microarray Data

In this study, we present a novel approach to improve cancer classification using high-dimensional microarray data. The proposed method combines a hybrid filter and a genetic algorithm-based feature selection process, incorporating Chi-square and Recursive Feature Elimination (RFE) techniques to identify critical gene expressions for cancer classification. Experiments using diverse datasets have yielded significant results. In the Lung Cancer Dataset, Logistic Regression Analysis (LR) and Support Vector Machine (SVM) achieved remarkable accuracy rates of 97.56%, with a precision and recall of 98.0%, resulting in an F1-score of 97.0%. This highlights the effectiveness of the feature selection method in enhancing classification accuracy. In the Ovarian Cancer Dataset, Gradient Boosting emerged as the top-performing classifier, achieving an accuracy of 92.85% along with precision, recall, and F1-score values of 94.0%, 93.0%, and 92.0%, respectively. These results demonstrate the versatility of the proposed feature-selection approach. This demonstrates the adaptability of the proposed feature selection technique in improving classifier performance. In summary, the hybrid filter and genetic algorithm-based feature selection method, incorporating Chi-square and RFE, proved to be a valuable tool for enhancing cancer classification in high-dimensional microarray data. The consistently high accuracy, precision, recall, and F1-score across diverse cancer datasets underscore the effectiveness and versatility of the proposed approach, holding promise for the development of more accurate cancer classification models in the future.