An Empirical Comparison of Classification Machine Learning Models Using Medical Datasets

Abstract

Classification is a supervised learning model where the class labels are accurately identified for future samples. Medical data is an important source for understanding and improving health outcomes and classification algorithms are often used to analyze these data. Learning models give significant experiences into the situational needs of patients. Various hypotheses have been carried out on different datasets yet it is truly challenging to track down which model is suitable. Proposed work compares the performance of classification models like LR, DT, SVM, NB, KNN, and RF on various datasets. SVM classifier yields accuracy of 0.59 for the Diabetic dataset as it considers individual model opinion, while RF classifier surpassed them both with accuracy 0.9974 for the breast cancer Wisconsin dataset since it is an ensemble approach that takes majority opinions. These findings highlight the need for careful consideration of the choice of classification model when analyzing medical data and provide valuable insights for researchers and practitioners working with these data.