Customer Churn Prediction on Structured Data Using FT-Transformer and Stacking Ensembles

Customer churn prediction is essential across data-driven industries such as insurance, digital banking, e-commerce, and subscription platforms, where retaining existing customers usually incurs lower cost than acquiring new ones. Churn prediction on structured tabular datasets remains challenging due to class imbalance, nonlinear feature interactions, and heterogeneous feature types. Tree-based ensembles continue to demonstrate strong performance in these settings, often outperforming conventional neural networks. This study introduces a validated hybrid architecture that integrates feature-tokenized transformers (FT- Transformer) with gradient-boosted trees using calibration-aware stacking. The framework addresses gaps in statistical validation, probability calibration, and reproducibility that persist in existing work. An FT-Transformer models higher-order feature interactions via self-attention, while XGBoost captures gradientboosted decision boundaries with complementary inductive biases. The models are ensembled using out-of-fold (OOF) stacking with a logistic regression meta-learner to improve generalization under class imbalance. On a public bank churn dataset (10,000 customers, 20% churn rate), the hybrid model achieves 62.10% F1, 0.861 AUC-ROC, and 0.647 PR-AUC, outperforming the Multi-Layer Perceptron (MLP) baseline by 3.37 F1 points (p < 0.001) and 0.027 AUC under 5x5 cross-validation. Ablation studies confirm that both the transformer component and stacking strategy contribute materially to performance. The proposed methodology provides a reproducible and extensible reference architecture for modern churn prediction on structured tabular data, bridging recent advances in attention-based modeling with ensemble techniques.