Brain tumors represent a severe category of neurological disorders, with a complex prediction task that requires an accurate and early detection for efficient treatment planning. Magnetic Resonance Imaging (MRI) represents the popular modality utilized in brain tumors diagnostic; however, manual handling impacts the efficiency of prediction and results in delays. Automated techniques such as machine learning (ML) have revolutionized healthcare by enhancing the efficiency and accuracy of predicting critical medical conditions. This paper aims to automate brain tumors detection and classification based on ML and DL models, and an MRI dataset of 409 images. This research paper demonstrates the robustness of ML and DL techniques and attains an accuracy of 90% in ML methods and 97.6% in DL models.