Nowadays, in the health sector, disease prediction and classification are effectively possible with the right application of specific data mining and machine learning (ML) techniques. On the other hand, because artificial diagnostic methods are not very accurate, medical experts are not able to properly incorporate them into the process of diagnosing brain diseases. In this work, we have focused on an efficient and automated classification technique known as the Hyper-Parameter SVM model (HP-SVM) for classifying different brain tumor (BT) diseases by using magnetic resonance (MRI) images. Data augmentation and transfer learning have enhanced the model's categorization performance. To achieve the quick diagnosis of various features to be extracted from the MRI images, the HP-SVM model gives a better result with 99.69% accuracy in comparison to other standard techniques. The proposed model can be helpful for proper diagnosis, and that can save the lives of people. Given the good predictive findings, we believe that the suggested model has the potential to be utilized for the diagnosis of brain tumors in Internet of Things (IoT) healthcare systems.