Niğde Ömer Halisdemir Üniversitesi Mühendislik Bilimleri Dergisi
Yazarlar: İlknur BÜTÜNER, Burak KAPLAN, Kemal ADEM
Konular:Bilgisayar Bilimleri, Bilgi Sistemleri
DOI:10.28948/ngumuh.655720
Anahtar Kelimeler:Machine learning,Diagnosis of Parkinson's,RseslibKnn,A1DE,KNN
Özet: Parkinson's disease is a neurological disease that affects the quality of life of people. Parkinson's disease is a disease that negatively affects the central nervous system. It can lead to death of patients. Therefore, early detection of Parkinson's disease is extremely important. Symptoms of Parkinson's disease can be detected with computer-assisted diagnostic systems based on potentially advanced machine learning techniques. In this study, kNN, RseslibKnn and A1DE machine learning methods were used for the diagnosis of Parkinson's disease. The aim of the study was to compare the success rates of the machine learning methods on the Parkinson's disease dataset and to present the most appropriate decision support system. The data set was composed of 252 samples and 753 attributes from the ‘UC Irvine Machine Learning Repository’ database. Different studies on the literature are also examined and compared. Experimental studies were conducted on different cross-validations and the average of these studies was given as a result of success. At the end of the study, the Parkinson's disease data set was classified with kNN, RseslibKnn and A1DE machine learning methods and then the training and test results were evaluated based on accuracy, sensitivity and specificity values. When all the methods examined with different cross validity values were examined, RseslibKnn method gave the highest success result with an average of %97,61 accuracy rate. As a result of the evaluation, the recommendations of the RseslibKnn machine learning method on decision support systems for the detection of Parkinson's disease were made.