Machine learning is increasingly used across safety-critical domains. In this work, explainable machine learning is applied to predict Remaining Useful Life (RUL) of turbofan engines using the NASA C-MAPSS dataset. Multivariate telemetry is transformed into history-aware features (10-cycle rolling means/standard deviations and first differences) together with a cycle-normalized age signal. Four regressors—Random Forest, XGBoost, LightGBM, and Support Vector Regression—are trained and then combined via a stacked ensemble. The learning algorithms and key hyperparameters are outlined, and models are evaluated using Mean Squared Error (MSE; absolute error magnitude), R² (explained variance), Mean Absolute Percentage Error (MAPE; average relative error), and Symmetric MAPE (sMAPE; scale-free percentage error) on an engine-wise 80/20 split. Quantitative results identify the stacked model as best overall, with LightGBM and Random Forest as strong single learners. Qualitative analysis employs SHapley Additive exPlanations (SHAP) to rank attributes that influence RUL, emphasizing life-cycle progression and a compact set of windowed sensor statistics. The paper closes with practical implications for maintenance scheduling

Keywords: Remaining Useful Life (RUL); turbofan engines; Random Forest (RF); XGBoost; LightGBM (LGBM); Support Vector Regression (SVR); stacking; SHapley Additive exPlanations (SHAP); explainable AI