Answers are grounded in this paper and cite their sources.
This study proposes a profit-based ensemble model with integrated threshold optimization for hotel booking cancellation prediction. The proposed framework proposes a deployable intervention rule by directly maximizing the expected maximum profit criterion (EMPC) under a realistic payoff structure. Two base learners are trained using a cross-validation protocol to generate probability estimates. These risk estimates along with a single contextual engagement signal are integrated within the decision layer to produce an intervention score. A two-stage evolutionary learning strategy employing differential evolution is developed. The first stage optimizes a smooth surrogate of the EMPC to learn profit-aware ensemble weights, while the second stage selects the operating threshold by maximizing the EMPC. The model achieved EMPC of € 26.60 and area under the curve of 0.9288. Furthermore, our framework ensures explainability through feature importance analysis and interpretation of the decision-layer weights. The proposed framework offers a decision support system for hotel cancellation management.
No takes yet. Share an insight, caveat, or question.
Hosny et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: