Abstract—The severity of ransomware threats is increasing significantly, posing substantial risks to individuals and organizations. To address this problem, several techniques have been suggested; however, they are mostly signaturebased, capable of identifying conventional ransomware but inadequate in detecting novel variants. This paper introduces a ransomware detection framework using distributed gradient boosting and explainable artificial intelligence. The suggested design utilizes the XGBoost algorithm for classification and explainable artificial intelligence for feature importance. The proposed design is deployed and tested using a dataset of 84 features. Weight-based feature selection is used to minimize the number of features. The selected features are used to train the XGBoost classification algorithm. Furthermore, the significance of the chosen features is assessed using SHapley Additive exPlantions (SHAP). The results show better performance in terms of accuracy, precision, recall, and F1-Score. The suggested approach outperforms the baseline machine learning algorithms and comparable results against state-of-the-art models across all performance criteria used.