Parkinson’s disease (PD) primarily affects the motor abilities as well as the neurological system of a person in the long term as it is a degenerative disease. Some of the initial symptoms are stiffness of muscles and an unsteady gait. At this stage, difficulties in walking do not manifest obviously. Moreover, a preliminary diagnosis based on scans or blood tests is generally inconclusive. Hence, identification of the onset of PD is a challenge for medical practitioners. Slurring of speech, however, does act as a red flag in the early prediction of PD. This study uses samples of speech taken from patients of PD as well as normal people for predicting PD. This dataset was generated by Max Little from the University of Oxford in collaboration with the National Centre for Voice and Speech in Denver, Colorado, through the recording of gathered speech signals. Even though clinical exams factor in a large volume of data with various features, doctors still find it difficult to decide whether a person is suffering from PD or not, on the basis of what type of data they have. The system proposed in this paper makes use of different classifiers used, including Gaussian Naive Bayes, K-nearest neighbors, logistic regression, decision trees and support vector machines. Among all the classifiers used, decision trees achieved the greatest accuracy rate, at 89.74% and also dominant voice features analysed.