Performance Analysis of Gaussian Naïve Bayes for Cancer Classification on RNA Metabric Mutation Data Based on Smoothing Parameter

Breast cancer is the second most prevalent cause of cancer-related deaths among women, behind lung cancer. It develops in the breast cells and occurs when these cells begin to grow abnormally and uncontrollably. Machine learning can improve breast cancer early detection. This study uses a number of machine learning models, including Support Vector Machine (SVM), Naïve Bayes, and K-Nearest Neighbor (KNN), to analyze breast cancer datasets and assess performance. The use of Naïve Bayes in breast cancer detection is the primary focus of this research with the GaussianNB variant on the smooth parameter. The dataset used in this paper is METABRIC RNA Mutation, which consists of 1,904 samples and 692 features. To enhance model performance in predicting breast cancer types, Features are chosen using kernel-based techniques for numerical features Using the categorical chi-square test features. The dataset's breast cancer type test results indicate that Naïve Bayes has the highest performance at 97% by utilizing the var smoothing parameter, while Support Vector Machine (SVM) accuracy is 92%, while K-Nearest Neighbor (KNN) accuracy is 95%. Because it can accurately employ the characteristics chosen using the kernel-based technique and chi-square test, Naïve Bayes is the best model for predicting the kinds of breast cancer.