This study presents an intelligent system that uses ensemble learning techniques to predict Parkinson disease. To train the model, data gathered from Kaggle is utilized. By selecting features based on mutual information, the prediction model's accuracy and efficacy are increased. The work classifies Parkinson illness using the Adaboost and XGBoost algorithms. The recommended methodology demonstrated exceptional prediction performance in the findings, achieving an even greater accuracy rate of 95.87% with XGBoost and 87.18% accuracy rate using Adaboost. Through the efficient integration of classification algorithms with feature selection techniques, this work marks a significant breakthrough in the use of machine learning to medical diagnosis and prognosis.