The Journal of Financial Data Science· 2026Q2
Stability-Weighted Ensemble Feature Importance for Financial Applications
- 2citations
- Q2SCImago
- 2026year
Short summary
A new method, SWEFI, combines stability selection with ensemble learning to provide stable and accurate feature importance scores for noisy financial data, outperforming existing methods in experiments.
AI-generated from the title and abstract; the full text is not read.
Key points
- Introduces SWEFI, a novel method combining stability selection and ensemble learning for feature importance.
- Addresses challenges of noisy and non-stationary financial data, which hinder traditional methods.
- SWEFI assesses feature relevance using multiple techniques and learning models.
- Experimental results demonstrate SWEFI's superior stability and accuracy over existing methods.
AI-generated from the title and abstract; the full text is not read.
Abstract
Robust feature importance is crucial in financial applications to enhance model interpretability and performance. Traditional methods often falter when applied to noisy, non-stationary financial data, leading to inconsistent and unreliable feature importance scores. This paper introduces the stability-weighted ensemble Feature Importance (SWEFI) method, which combines stability selection with ensemble learning to address these challenges. SWEFI aims to deliver stable and interpretable feature importance scores, thereby improving model reliability in high-stakes financial environments. The methodology integrates multiple feature importance techniques and learning models, assessing feature relevance through a comprehensive and systematic approach. Extensive experiments using synthetic and real-world financial datasets demonstrate SWEFI's superior stability and accuracy compared to existing methods. This research underscores the importance of stable feature importance measures, facilitating better decision-making and enhancing trust in machine learning models.
The authors' abstract, as published at the source. The Journal of Financial Data Science, 2026 · DOI ↗
Continue with a free account
Ask the paper: 3 free questions a day about this paper; save it, get its citation, new summaries every day for your field. Takeaways are Premium.
Continue free on the webSign in with Google or Apple; no card needed. You come back to this paper.
On your phone:
Field: Artificial Intelligence
Artificial IntelligenceComputer Science