This study aims to improve the prognosis of gliomas by employing a combination of Machine Learning (ML) and Deep Learning (DL) techniques, with a specific focus on classifying glioma grades into Low-Grade Gliomas (LGG) and Glioblastoma Multiforme (GBM). To achieve high classification accuracy, the research integrates a comprehensive ensemble of models, including Random Forest, Decision Tree, Logistic Regression, K-Nearest Neighbors (KNN), AdaBoost, Support Vector Machine (SVM), CatBoost, LightGBM, XGBoost, Artificial Neural Network (ANN), and Convolutional Neural Network (CNN). Additionally, to broaden the comparative analysis, the study incorporates Bernoulli Naive Bayes (BernoulliNB), Linear Discriminant Analysis (LDA), and Passive Aggressive Classifier, enhancing the diversity of learning paradigms evaluated. To improve learning efficiency and reduce model complexity, K-Best Feature Selection is applied as a preprocessing step to identify the most relevant features contributing to glioma grade classification. Furthermore, the study emphasizes model transparency and clinical interpretability, integrating Explainable AI (XAI) techniques—particularly the SHapley Additive exPlanations (SHAP) algorithm. SHAP provides meaningful insights into feature importance and model behavior, empowering clinicians to make informed, data-driven decisions.