Parkinson's disease (PD) and essential tremor (ET) are two kinds of tremor disorders which always confusing doctors in clinical diagnosis. Early experiments on structural MRI have already shown that Parkinson's disease can cause pathological changes in the brain region named Caudate-R (a part of Basal ganglia) while essential tremor cannot. Although there are many research work on the classification of PD and ET, they didn't achieve the automatic classification of the two diseases. But big data brings new opportunities to the classification of PD and ET. In order to achieve this, we proposed a machine learning framework which contains two parts: feature extraction and classification to achieve the classification of Parkinson's disease and Essential Tremor. At first, we used principal component analysis (PCA) and Kernel PCA to extract discriminative features from structural MRI data, respectively. Then SVM classifier is employed to classify PD and ET. We used statistical analysis and machine learning method to test the differences between PD and ET in specific brain regions. As a result, the machine learning method has a better performance in extracting the differential brain regions. The highest classification accuracy is up to 93.75% in the differential brain regions.