The pipeline for machine learning often contains cross-stage defects, such as split leakage, missing values, feature corruption and label noise; isolated checks cannot reliably trace the origin or effect downstream. Methods: We propose DAPA, a defect-aware pipeline agent that integrates data contracts, provenance graphs, robust statistical sensing, cross-validated label auditing, constrained repair planning, counterfactual evaluation and an immutable audit ledger. A framework was used to conduct controlled defect injection on the three public datasets for twenty stratified repetitions. Results: DAPA had a pooled Macro-F1 of 0.9366, improved the defective pipeline by 0.0203, and reduced the expected calibration error from 0.1540 to 0.0486. The paired Wilcoxon test was still significant after Holm correction (p=0.0012). Leakage and missingness were detected perfectly under the controlled protocol, whereas label-noise recall was 0.4894. In short, defect-aware orchestration has improved the prediction stability and traceability of models, but problems of uncertain label repair and high-dimensional search still exist in all areas.

Yue Cheng. Research on Semantic Analysis and Intelligent Attribution of Software Defect Reports Based on Natural Language Processing. International Journal of Big Data Intelligent Technology (2026), Vol. 7, Issue 3: 9-19. https://doi.org/10.38007/IJBDIT.2026.070302.

[1] Chai C, Wang J, Luo Y, Niu Z, Li G. Data Management for Machine Learning: A Survey. IEEE Transactions on Knowledge and Data Engineering, 2023, 35(5): 4646-4667. https://doi.org/10.1109/TKDE.2022.3148237

[2] Kapoor S, Narayanan A. Leakage and the reproducibility crisis in machine-learning-based science. Patterns, 2023, 4(9): 100804. https://doi.org/10.1016/j.patter.2023.100804

[3] Mazumder M, Banbury C, Yao X, et al. DataPerf: Benchmarks for Data-Centric AI Development. Advances in Neural Information Processing Systems, 2023, 36: 5320-5347. https://proceedings.neurips.cc/paper_files/paper/2023/hash/112db88215e25b3ae2750e9eefcded94-Abstract-Datasets_and_Benchmarks.html

[4] Jiang KF, Liang W, Zou J, Kwon Y. OpenDataVal: A Unified Benchmark for Data Valuation. Advances in Neural Information Processing Systems, 2023, 36: 28624-28647. https://proceedings.neurips.cc/paper_files/paper/2023/file/5b047c7d862059a5df623c1ce2982fca-Paper-Datasets_and_Benchmarks.pdf

[5] Goswami M, Challu C, Callot L, Minorics L, Kan A. AQuA: A Benchmarking Tool for Label Quality Assessment. Advances in Neural Information Processing Systems, 2023, 36. https://proceedings.neurips.cc/paper_files/paper/2023/file/fc20ea8d104cab737a5561096f9bde9b-Paper-Datasets_and_Benchmarks.pdf

[6] Siddiqi S, Kern R, Boehm M. SAGA: A Scalable Framework for Optimizing Data Cleaning Pipelines for Machine Learning Applications. Proceedings of the ACM on Management of Data, 2023, 1(3): 218:1-218:26. https://doi.org/10.1145/3617338

[7] Jäger S, Biessmann F. From Data Imputation to Data Cleaning—Automated Cleaning of Tabular Data Improves Downstream Predictive Performance. Proceedings of AISTATS, PMLR 238, 2024: 3394-3402. https://proceedings.mlr.press/v238/jager24a.html

[8] Guha S, Khan FA, Stoyanovich J, Schelter S. Automated Data Cleaning Can Hurt Fairness in Machine Learning-Based Decision Making. IEEE Transactions on Knowledge and Data Engineering, 2024, 36(12): 7368-7379. https://doi.org/10.1109/TKDE.2024.3365524

[9] Rosenblatt M, Tejavibulya L, Jiang R, Noble S, Scheinost D. Data Leakage Inflates Prediction Performance in Connectome-Based Machine Learning Models. Nature Communications, 2024, 15: 1829. https://doi.org/10.1038/s41467-024-46150-w

[10] Tschalzev A, Marton S, Lüdtke S, Bartelt C, Stuckenschmidt H. A Data-Centric Perspective on Evaluating Machine Learning Models for Tabular Data. Advances in Neural Information Processing Systems, 2024, 37: 95896-95930.

[11] Gao H, Cai S, Dinh T T A, Huang Z, Ooi B C. CtxPipe: Context-Aware Data Preparation Pipeline Construction for Machine Learning. Proceedings of the ACM on Management of Data, 2024, 2(6): 1-27. https://doi.org/10.1145/3698831

[12] Bayram F, Ahmed B S. Towards Trustworthy Machine Learning in Production: An Overview of the Robustness in MLOps Approach. ACM Computing Surveys, 2025, 57(5): 1-35. https://doi.org/10.1145/3708497

[13] Alturayeif N, Hassine J. Data Leakage Detection in Machine Learning Code: Transfer Learning, Active Learning, or Low-Shot Prompting? PeerJ Computer Science, 2025, 11: e2730. https://doi.org/10.7717/peerj-cs.2730

[14] Chan JS, Chowdhury N, Jaffe O, et al. MLE-bench: Evaluating Machine Learning Agents on Machine Learning Engineering. International Conference on Learning Representations, 2025. https://openreview.net/forum?id=6s5uXNWGIh

[15] Trirat P, Jeong W, Hwang S J. AutoML-Agent: A Multi-Agent LLM Framework for Full-Pipeline AutoML. Proceedings of the 42nd International Conference on Machine Learning, PMLR 267, 2025: 60099-60146. https://proceedings.mlr.press/v267/trirat25a.html

[16] Wang, Z. (2026). Research on Operation Data Analysis and Efficiency Improvement of the New Energy Industry Based on Machine Learning. Procedia Computer Science, 285, 605-612.

[17] Wang, Z. (2026). Construction of a Data-driven Decision-making Framework for Bulk Commodity Supply Chains. Engineering Advances, 6(3).