Parkinson is a disease related to brain disorder which leads to inadvertent movements like stiffness, shaking and unsteadiness. The Parkinson disease symptoms generally take up slowly and get worsen over period of time. It leads to difficulty in talking and walking by an individual. The main genesis of Parkinson Disease is the death or impairedness of basal ganglia nerve cell that controls the movement of and individual. Our dataset of Parkinson’s is taken from UCI Repository and the dataset is generated in association with National Center For Voice And Speech Colorado and Max Lattice of Oxford University. The dataset consist of biomedical voice measurement of 31 people. Data Preprocessing is performed including outlier removal and normalization, followed by feature Selection using Principle Component Analysi, and then machine learning algorithms like Logistic Regression, Support Vector Machine, Naive Bayes and Random Forest are applied for Parkinson Classification. Here The Random Forest outperforms other machine learning algorithms with 94.4% accuracy. Additionally threshold analysis is conducted within the Random Forest Model, Optimizing the split decision for noisy biomedical data. Overall feature importance analysis is also performed for all four algorithms.