Moot: A repository of many Multi-objective optimization tasks

MOOT (Multi Objective Optimization Tasks) is a curated repository of real-world multi-objective optimization tasks drawn from recent software engineering and systems research. These tasks span software configuration, cloud and systems tuning, project health, process modeling, and hyperparameter optimization, among others.

The goal of MOOT is to provide a reusable, extensible benchmark suite that enables:

• Rigorous comparison of optimization algorithms across many tasks.
• Replication and extension of prior empirical studies in search-based software engineering.
• Rapid exploration of new ideas in optimization, meta-learning, and automated software analytics.

MOOT at a glance

# Datasets	Dataset Type	File Names	Primary Objective	x/y	# Rows	Cited By
25	Specific Software Configurations	SS-A to SS-X, billing10k	Optimize software system settings	3-88/2-3	197–86,059	[1,2,3,4,5,6,7,8,9,10]
12	PromiseTune Software Configurations	7z, BDBC, HSQLDB, LLVM, PostgreSQL, dconvert, deeparch, exastencils, javagc, redis, storm, x264	Software performance optimization	9-35/1	864-166,975	[11,12,13,14,15]
1	Cloud	HSMGP num	Hazardous Software Management Program data	14/1	3,457	[1,2,3,4,5,12]
1	Cloud	Apache AllMeasurements	Apache server performance optimization	9/1	192	[1,2,3,4,5,12]
1	Cloud	SQL AllMeasurements	SQL database tuning	39/1	4,654	[1,2,3,4,5]
1	Cloud	X264 AllMeasurements	Video encoding optimization	16/1	1,153	[1,2,3,4,5]
7	Cloud	(rs—sol—wc)*	Misc configuration tasks	3-6/1	196–3,840	[1,2,3,4,5,6,7]
35	Software Project Health	Health-ClosedIssues, -PRs, -Commits	Predict project health and developer activity	5/2-3	10,001	[1,2,3,4,5,6,20]
3	Scrum	Scrum1k, Scrum10k, Scrum100k	Configurations of the scrum feature model	124/3	1,001–100,001	[1,2,4,6]
8	Feature Models	FFM-, FM-	Optimize number of variables, constraints and Clause/Constraint ratio	128-1,044/3	10,001	[1,2,4,6]
1	Software Process Model	nasa93dem	Optimize effort, defects, time and LOC	24/3	93	[2,5,6,20]
1	Software Process Model	COC1000	Optimize risk, effort, analyst experience, etc	20/5	1,001	[1,2,5,20,21]
4	Software Process Model	POM3 (A–D)	Balancing idle rates, completion rates and cost	9/3	501–20,001	[2,4,5,6,20]
4	Software Process Model	XOMO (Flight, Ground, OSP)	Optimizing risk, effort, defects, and time	27/4	10,001	[2,4,5,6,20,21]
3	Miscellaneous	auto93, Car_price, Wine_quality	Miscellaneous	5-38/2-5	205–1,600	[1,2,3,4,5,6,20]
4	Behavioral	all_players, student_dropout, HR-employeeAttrition, player_statistics	Analyze and predict behavioral patterns	26-55/1-3	82–17,738	[29,31,32,33]
4	Financial	BankChurners, home_data, Loan, Telco-Churn	Financial analysis and prediction	19-77/2-5	1,460–20,000	[34,35,36,37]
3	Human Health Data	COVID19, Life_Expectancy, hospital_Readmissions	Health-related analysis and prediction	20-64/1-3	2,938–25,000	[38,39,40]
2	Reinforcement Learning	A2C_Acrobot, A2C_CartPole	Reinforcement learning tasks	9-11/3-4	224–318	—
5	Sales	accessories, dress-up, Marketing_Analytics, socks, wallpaper	Sales analysis and prediction	14-31/1-8	247–2,206	[30,41,42]
2	Software testing	test120, test600	Optimize the class	9/1	5,161	—
127	Total

Note: "x/y" denotes the number of independent and dependent attributes. All citations are detailed in docs/cited_by.md.

Team

MOOT is a collaborative effort between researchers at North Carolina State University and the University of Birmingham led by Dr. Tim Menzies and Dr. Tao Chen to curate and maintain a large, open repository of multi-objective optimization tasks for software engineering and related domains. Building on prior experience with large-scale empirical resources such as the PROMISE repository and extensive work in search-based software engineering and optimization, the MOOT team offers a robust foundation for rigorous, reproducible research.

Citing MOOT

If you use MOOT in your research, please cite the following paper:

Menzies, T., Chen, T., Ye, Y., Ganguly, 
K. K., Rayegan, A., Srinivasan, S., & Lustosa, A. (2025). 
MOOT: a Repository of Many Multi-Objective Optimization Tasks. 
arXiv:2511.16882

@misc{menzies2025mootrepositorymultiobjectiveoptimization,
      title={MOOT: a Repository of Many Multi-Objective Optimization Tasks}, 
      author={Tim Menzies and Tao Chen and Yulong Ye and Kishan Kumar Ganguly 
              and Amirali Rayegan and Srinath Srinivasan and Andre Lustosa},
      year={2025},
      eprint={2511.16882},
      archivePrefix={arXiv},
      primaryClass={cs.SE},
      url={https://arxiv.org/abs/2511.16882}, 
}

Key Resources

• MOOT GitHub repository: https://github.com/timm/moot
• MOOT paper: “MOOT: a Repository of Many Multi-Objective Optimization Tasks” – https://arxiv.org/abs/2511.16882
• Tim Menzies’ homepage: https://timm.fyi
• “The Case for Compact AI” (Communications of the ACM article discussing compact AI and related ideas): https://cacm.acm.org/opinion/the-case-for-compact-ai/

Moot is such a good name for datasets to be used to assess different algorithms. Its definition is

(noun) a mock trial set up to examine a hypothetical case as an academic exercise. "the object of a moot is to provide practice in developing an argument"