Biological research is becoming increasingly database driven and statistical learning can be used to discover patterns in the biological data. In the thesis, the supervised learning approaches are utilized to analyze the Oxford Parkinson’s disease detection data and build models for prediction or classification. We construct predictive models based on training set, evaluate their performance by applying these models to an independent test set, and find the best methods for predicting whether people have Parkinson’s disease. The proposed artificial neural network procedure outperforms with the best and highest prediction accuracy, while the logistic and probit regressions are preferred statistical models which can offer better interpretation with the higher prediction accuracy compared to other proposed data mining approaches.