Akademik Veri Yönetim Sistemi
Balıkesir Üniversitesi Fen Bilimleri Enstitüsü Dergisi, cilt.27, sa.1, ss.269-281, 2025 (TRDizin)
In this study, artificial intelligence was applied to accurately diagnose and predict erythemato-squamous diseases (ESDs). Feature selection was performed for 34 features in the dataset with the wrapper feature selection method. 18 features were selected using the feature selection method. In the analyses performed with machine learning algorithms, results were obtained and compared with both the initial 34 features and the selected 18 features. Six different machine learning classification algorithms were compared for erythemato-squamous diseases. Naive Bayes algorithm was determined as the most successful algorithm in the diagnosis and prediction of erythemato-squamous diseases with an accuracy rate of 99.45%. In addition, it was determined that the applied feature selection method increased the performance of all algorithms. When the results obtained in the study are examined, it is seen that wrapper feature selection plays an important role in improving the performance of machine learning models.