Accurate short-term load forecasting is essential for reliable power system operation, particularly under the increasing uncertainty caused by abnormal weather and socio-economic fluctuations. This study presents a month-conditioned boosting framework that integrates SHapley Additive Explanations (SHAPs) into model refinement. A baseline XGBoost model was first compared with linear and tree-based regressors, followed by enhancements through lagged and rolling-window features as well as loss weighting for vulnerable months. To further improve the performance, SHAP analysis was employed to identify the dominant error-contributing features, which guided the construction of targeted month-specific interaction terms for retraining. Experimental results based on rolling-origin cross-validation showed that this approach significantly reduced the RMSE and MAPE, particularly during high-variance summer months. Moreover, the SHAP interpretation revealed the varying roles of seasonal demand structures and socio-economic mobility, thereby enhancing transparency and operational insight. The proposed framework demonstrated that embedding explainability into the learning loop improved predictive accuracy and ensured interpretability, offering a data-driven solution for electricity demand forecasting in practical settings.

Trustworthy analytics for healthcare require models that are not only accurate but also interpretable and robust under distributional perturbations. In this paper, we propose an interpretable stacked ensemble framework that repurposes TabNet from an end-to-end classifier into an attention-guided feature generator for downstream learners. We constructed a dual-channel stacking architecture in which TabNet-derived embeddings and original tabular features were fed into heterogeneous gradient-boosted base learners (XGBoost and LightGBM) to enhance representation diversity, and were integrated by an interpretable logistic-regression meta-learner. For transparent and unbiased evaluation, we employed nested stratified cross-validation with fixed-budget hyperparameter tuning with systematic ablation studies. Experiments on the public Wisconsin Diagnostic Breast Cancer dataset showed that the proposed model achieves strong and stable performance (average accuracy 97.8% ± 1.0% under nested cross-validation) compared to a single TabNet baseline and conventional ensemble variants. Moreover, we assessed robustness under out-of-distribution-style covariate perturbations by injecting Gaussian noise at varying intensities, demonstrating that the stacking design mitigates the noise sensitivity of TabNet-derived representations and maintains a balanced sensitivity–specificity trade-off via adaptive thresholding. Overall, the proposed framework provides a reproducible template for combining deep tabular representation learning with explainable ensemble decision-making toward reliable data science applications in high-stakes domains.