Our research aimed to automate brain tumor patients survival rate prediction using machine learning models. We utilized the BRATS 2020 database, which provides segmented MRI scans and clinical data of 236 patients. Due to the complexity of the problem and limited data, we transformed it into a classification problem with categorical outputs, allowing us to use simpler models. After fine-tuning various algorithms, we evaluated their performance using accuracy, F1 score, sensitivity, and precision metrics. Our results showed that decision trees and the voting classifier achieved the highest overall performance, while non-linear support vector machines performed well. K-nearest neighbors showed high sensitivity and precision values, while Gaussian naive Bayes had the lowest performance overall. Our findings suggest that classification algorithms can accurately predict the overall survival of brain tumor patients. However, it’s important to note that this approach cannot replace the regression approach, which can determine the exact number of days left to the patient. To further improve our results and potentially merge the two methods, we need more data to develop even more accurate classification algorithms. Among our tested models, the decision trees and voting classifier stood out as the top performers, with accuracy scores 0.76. Meanwhile, non-linear SVM and SVM polynomial models also showed strong performance, with accuracy scores of 0.69 and 0.70, respectively. K-nearest neighbors with 27 neighbors achieved an impressive accuracy score of 0.72, and logistic regression showed moderate performance with an accuracy score of 0.63. Our research highlights the potential of machine learning to assist in diagnosing and treating brain tumor patients, and we hope our findings can contribute to future advancements in this field.