A Comparison of Traditional and Ensemble Machine Learning Approaches for Parkinson’s Disease Classification

Abstract

Parkinson’s disease (PD) is a movement-related disorder that negatively impacts the central nervous system. It is progressive, which means that patients’ condition worsens over time. According to literature, it is more common among men as compared to women. It affects a person in various ways such as in its initial stages, it may cause a few tremors in certain body parts and the consequences worsen throughout its progression—difficulties in speech, writing, and movement all develop eventually. This paper performs an effective comparison of an individual as well as ensemble machine learning methods on quantitative acoustic measure data for PD classification. In individual models, category Support vector machine (SVM), Naive Bayes, Decision Tree, K-Nearest Neighbor (KNN), and Logistic Regression approaches have been adopted. In the category of ensemble models Random Forest, Gradient boosting, Adaptive Boosting, and XGBoost. The models were evaluated and compared using performance metrics such as Recall, F1 score, Precision, and Accuracy. Results indicate that XGBoost performs well than others.