Brain tumors are life-threatening conditions that require accurate diagnosis for effective treatment, and magnetic Resonance Imaging plays a significant role in the diagnosis of brain tumors.  Categorization of the tumor type is essential for making necessary medical decisions. Brain Tumor is commonly classified as Normal, Benign, or Malignant. There is a publicly available dataset in Kaggle for brain tumor classification with classes such as meningioma, pituitary gland, glioma, and no tumor. In our work, we proposed a novel method, Smote Tomek with Support Vector Machine (SVM), for brain tumor classification.The Combined sampling technique of smote from oversampling and Tomek from Undersampling was applied to compensate for the imbalance in the data.First, we implemented the combined technique of SMOTETomek to clear this data imbalance, leading to an improvement in the results.We then fitted the balanced data to the SVM classifier. Hence, our proposed method produces the best result with an accuracy of 95% for categorizing the data as pituitary tumor or no tumor. It also provides better results in terms of other metrics such as sensitivity and specificity. This method was also compared with other competent classifiers and was found to be an effective method for the classification of brain tumor data.