Integrating Data Smell Identification And Mitigation Techniques: A Methodical Approach To Data Quality
DOI:
https://doi.org/10.29284/x1g61p81Keywords:
Data Smells, Logistic Regression, Decision Trees, Random Forests, XGBoost, Multi-Layer Perceptron.Abstract
Artificial intelligence (AI) system reliability mainly depends on the input data quality. However, subtle issues known as data smells can significantly degrade model performance and reliability. However small issues like data smells can deteriorate the performance and reliability of the model. This study inspects the impact of six categories of data smells -Believability, Consistency, Distribution, Redundancy, Understandability, and Miscellaneous - and evaluate their impact on five models: Logistic Regression, Decision Trees, Random Forests, XGBoost, and a Multilayer Perceptron (MLP), using the UCI Adult Income dataset. Distribution and Redundancy smells proved most harmful, reducing F1-scores by up to 75% for Logistic Regression and significantly affecting MLP, while ensemble methods showed higher strength. Consistency and other smells had smaller effects on accuracy but caused concerns for interpretability, ethics, and maintainability. This study introduces a Data Smell Severity Index to rank smell types with respect to impact, detection difficulty, occurrence, outline mitigation strategies and future directions like real-time validation and automated detection. Overall, this work provides an empirical foundation for addressing data smells and strengthening the credibility of AI systems.
References
1. David Sculley, Gary Holt, Daniel Golovin, Eugene Davydov, Todd Phillips, Diet mar Ebner, Vinay Chaudhary, Michael Young, Jean- Francois Crespo, and Dan Dennison. 2015. Hidden technical debt in machine learning systems. Advances in neural information processing systems 28 (2015).
2. H. Foidl, M. Felderer and R. Ramler, "Data Smells: Categories, Causes and Consequences, and Detection of Suspicious Data in AI-based Systems," 2022 IEEE/ACM 1st International Conference on AI Engineering – Software Engineering for AI (CAIN), Pittsburgh, PA,
3. USA, 2022, pp. 229-239, doi: 10.1145/3522664.3528590
4. G. Recupito, R. Rapacciuolo, D. Di Nucci and F. Palomba, "Unmasking Data Secrets: An Empirical Investigation into Data Smells and Their Impact on Data Quality," 2024 IEEE/ACM 3rd International Conference on AI Engineering – Software Engineering for AI (CAIN), Lisbon, Portugal, 2024, pp. 53-63.
5. Polyzotis, N., Zinkevich, M., Roy, S., Breck, E. and Whang, S., 2019. Data validation for machine learning. Proceedings of machine learning and systems, 1, pp.334-347.
6. Elouataoui, W., El Alaoui, I., El Mendili, S. and Gahi, Y., 2022. An advanced big data quality framework based on weighted metrics. Big Data and Cognitive Computing, 6(4), p.153.
7. V. Agrawal, “Data Validation : A Complex Challenge in Modern AI Systems,” International Scientific Journal of Engineering and Management, vol. 04, no. 01, pp. 1–9, Jan. 2025, doi: 10.55041/isjem01311.
8. S. E. Whang, Data Collection and Quality Challenges in Deep Learning: A Data-Centric AI
9. Perspective, VLDB Journal, 2023.
10. L. Fischer et al., AI System Engineering—Key Challenges and Lessons Learned, 2020.
11. H. Chen, Data Quality Evaluation and Improvement for Machine Learning, 2022.
12. T. Manju, Quality Challenges in Deep Learning Data Collection in Perspective of Artificial Intelligence, 2023.
13. V. Golendukhina, H. Foidl, M. Felderer, and R. Ramler, Preliminary findings on the occurrence and causes of data smells in a real-world business travel data processing pipeline, 2022.
14. H. Foidl and M. Felderer, Risk-based data validation in machine learning-based software systems, 2019.
15. D. Costal, C. Gómez, S. del Rey, and S. Martínez-Fernández, Using metrics for code smells of ML pipelines, 2024.
16. A. Torralba and A. A. Efros, Unbiased Look at Dataset Bias, CVPR, 2011.
17. A. Mikołajczyk and M. Grochowski, A Survey on Bias in Machine Learning Research, 2024.
18. M. Zaharia et al., Accelerating the Machine Learning Lifecycle with MLflow, IEEE Data Engineering Bulletin, 2018.
19. M. Hort et al., Bias Mitigation for Machine Learning Classifiers: A Comprehensive Survey, ACM Journal on Responsible Computing, 2023.
20. S. Siddique et al., Survey on Machine Learning Biases and Mitigation Techniques, Digital Journal, 2023.
21. T. Rekatsinas et al., HoloClean: Holistic Data Repairs with Probabilistic Inference, VLDB, 2017.
Downloads
Published
Issue
Section
License
This work is licensed under a Creative Commons Attribution 4.0 International License.
This work is licensed under a Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
