Exploring the Intersection of Metrology and Data Mining: A Review of Current Challenges and Future Directions

Data mining methods have significantly influenced various knowledge and market sectors. Improved, more complex, and flexible methods continue to be introduced. However, from a metrological perspective, these complex models may have high variance. Optimal model selection remains a challenge, often relying on heuristic techniques over solid knowledge. This manuscript reviews regression methods focusing on result validity, aiming to achieve comprehensive model characterization and uncertainty evaluation to ensure optimal and valid results for researchers and system users.