Classification algorithms are commonly employed in healthcare systems to aid decision support processes, such as treatment regimens, diagnosis, and illness prediction. The recent emergence of dominant variants of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), widely known as the coronavirus disease (COVID-19), has emphasized the significance of early detection for ensuring appropriate treatment and protecting unaffected populations. This study assesses the performance of various classification models on a COVID-19 dataset, utilizing two distinct feature selection methods: the wrapper method (WrapperSubsetEval) and the correlation-based feature subset evaluation (CfsSubsetEval). The effectiveness of these methods is evaluated based on the number of features selected for the reduced subset, execution time, and classifier accuracy. The experimentation is conducted using WEKA tools, and five different classifiers are selected for computation and comparison of accuracy: J48 decision tree (DT), support vector machine (SVM), naïve Bayes (NB), sequential minimal optimization (SMO), and k-nearest neighbor (KNN). The performance of each model is assessed using a 10-fold cross-validation technique, and the accuracy of the models is measured. The evaluation results, including comparisons before and after the implementation of the classification process and feature selection methods, indicate that KNN employing WrapperSubsetEval+KNN outperforms other algorithms, achieving the highest accuracy of 98.81%. In summary, the utilization of feature selection methods can be considered an effective approach for COVID-19 prediction.