Automating and Optimizing Data-Centric What-If Analyses on Native Machine Learning Pipelines

Summary: mlwhatif declaratively specifies data-centric what-if analyses over ML pipelines and auto-generates variants via patches. A 4-rule optimizer executes variants; instrumented dataflow plans enable linear speedups (up to 13x) and data-size independence. (summarized by gpt-5-nano on Feb 09 2026)

Paper ID: 6693
Venue: SIGMOD
Year: 2023
Pagerank: 5.6257796e-05
Overall Rank: 7,395 | 49.27%
DOI: 10.1145/3589273

Incoming Non-self Citations Over Time

Authors

1. Stefan Grafberger (University of Amsterdam)
2. Paul Groth (University of Amsterdam)
3. Sebastian Schelter (University of Amsterdam)

BibTeX Citation

@inproceedings{grafberger_sigmod23,
        title = {{Automating and Optimizing Data-Centric What-If Analyses on Native Machine Learning Pipelines}},
        author = {Grafberger, Stefan and Groth, Paul and Schelter, Sebastian},
        series = {{SIGMOD} '23},
        booktitle = {Proceedings of the {ACM} {SIGMOD} International Conference on Management of Data},
        publisher = {Association for Computing Machinery},
        doi = {10.1145/3589273},
        url = {https://dl.acm.org/doi/10.1145/3589273},
        year = {2023}
}

Incoming Citations (Sorted by Pagerank)

Showing 7 of 7 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank
6,614	Mitigating the Impedance Mismatch between Prediction Query Execution and Database Engine	2025	SIGMOD	5.8216658e-05
7,490	mlwhatif: What If You Could Stop Re-Implementing Your Machine Learning Pipeline Analyses Over and Over?	2023	VLDB	5.6061425e-05
8,949	Modyn: Data-Centric Machine Learning Pipeline Orchestration	2025	SIGMOD	5.3462965e-05
9,881	The Image Calculator: 10x Faster Image-AI Inference by Replacing JPEG with Self-designing Storage Format	2024	SIGMOD	5.2040783e-05
10,303	Understanding the Impact of Data Noise in Federated Learning: [Experiments & Analysis]	2026	SIGMOD	5.093636e-05
10,589	Morphing-based Compression for Data-centric ML Pipelines	2026	VLDB	5.093636e-05
11,043	mlidea: Interactively Improving ML Data Preparation Code via “Shadow Pipelines”	2025	VLDB	5.093636e-05

Outgoing Citations (Sorted by Pagerank)

Showing 23 of 23 cited papers.

Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.

Rank	Cited Paper	Year	Venue	Pagerank
103	DuckDB: an Embeddable Analytical Database	2019	SIGMOD	0.00034161428
128	Efficient and Extensible Algorithms for Multi Query Optimization	2000	SIGMOD	0.0003072825
420	Can Foundation Models Wrangle Your Data?	2023	VLDB	0.00018789852
582	ActiveClean: Interactive Data Cleaning For Statistical Modeling	2016	VLDB	0.00016148948
946	HoloDetect: Few-Shot Learning for Error Detection	2019	SIGMOD	0.00013054126
1,066	Efficient Task-Specific Data Valuation for Nearest Neighbor Algorithms	2019	VLDB	0.00012333161
1,340	Responsible Data Management	2020	VLDB	0.00011111667
1,431	Towards Scalable Dataframe Systems	2020	VLDB	0.00010807221
1,569	HELIX: Holistic Optimization for Accelerating Iterative Machine Learning	2019	VLDB	0.00010335423
1,756	SystemDS: A Declarative Machine Learning System for the End-to-End Data Science Lifecycle	2020	CIDR	9.8172465e-05
1,951	Interpretable Data-Based Explanations for Fairness Debugging	2022	SIGMOD	9.4252389e-05
1,952	Caravan: Provisioning for What-If Analysis	2013	CIDR	9.4233049e-05
2,316	Evaluating End-to-End Optimization for Data Analytics Applications in Weld	2018	VLDB	8.7596739e-05
2,593	Cost-Based Optimization of Decision Support Queries using Transient-Views	1998	SIGMOD	8.3649196e-05
2,657	Production Machine Learning Pipelines: Empirical Analysis and Optimization Opportunities	2021	SIGMOD	8.2887895e-05
2,865	End-to-end Optimization of Machine Learning Prediction Queries	2022	SIGMOD	8.0180243e-05
4,518	MLINSPECT: A Data Distribution Debugger for Machine Learning Pipelines	2021	SIGMOD	6.6474737e-05
4,805	Efficient Answering of Historical What-if Queries	2022	SIGMOD	6.5003306e-05
5,025	HYPER: Hypothetical Reasoning With What-If and How-To Queries Using a Probabilistic Causal Approach	2022	SIGMOD	6.3974766e-05
6,309	Materialization and Reuse Optimizations for Production Data Science Pipelines	2022	SIGMOD	5.9189554e-05
6,538	UPLIFT: Parallelization Strategies for Feature Transformations in Machine Learning Workloads	2022	VLDB	5.8477764e-05
8,859	Complaint-Driven Training Data Debugging at Interactive Speeds	2022	SIGMOD	5.356561e-05
11,509	Screening Native ML Pipelines with “ArgusEyes”	2022	CIDR	5.093636e-05

Semantically Similar Papers

#	Overall Rank	Paper	Year	Venue
1	5,704	Optimizing Data Pipelines for Machine Learning in Feature Stores	2023	VLDB
2	11,043	mlidea: Interactively Improving ML Data Preparation Code via “Shadow Pipelines”	2025	VLDB
3	4,518	MLINSPECT: A Data Distribution Debugger for Machine Learning Pipelines	2021	SIGMOD
4	11,512	Towards Observability for Machine Learning Pipelines	2022	CIDR
5	11,353	Reconstructing and Querying ML Pipeline Intermediates	2023	CIDR
6	2,657	Production Machine Learning Pipelines: Empirical Analysis and Optimization Opportunities	2021	SIGMOD
7	9,245	Towards Observability for Production Machine Learning Pipelines	2022	VLDB
8	6,309	Materialization and Reuse Optimizations for Production Data Science Pipelines	2022	SIGMOD
9	6,250	Lightweight Inspection of Data Preprocessing in Native Machine Learning Pipelines	2021	CIDR
10	7,490	mlwhatif: What If You Could Stop Re-Implementing Your Machine Learning Pipeline Analyses Over and Over?	2023	VLDB