Integration Of Feature Reduction Method With Feature Selection In Disease Prediction

Abstract

This research focuses on feature selection and evaluating the accuracy of machine learning models. The objective is to identify the most relevant features and optimize the model's performance while overcoming overfitting. The study begins by meticulously selecting 47 features out of an initial set of 132 features, which demonstrate improved accuracy. Several evaluation steps are conducted, including assessing feature coefficients, utilizing five model estimators, testing with ensemble methods, and evaluating results using the confusion matrix. The findings indicate that the applied technique successfully selects the most appropriate features and effectively mitigates overfitting. By considering feature coefficients, employing multiple model estimators, and leveraging ensemble techniques, the selected features significantly contribute to accurate predictions of the desired target. The evaluation results demonstrate high accuracy and the ability of the model to distinguish between positive and negative classes. Overall, the research showcases the effectiveness of the feature selection process and accuracy evaluation in optimizing machine learning models. The steps taken successfully address overfitting concerns and yield satisfactory outcomes. The study provides valuable insights into feature selection and model optimization for future research.