Missing Data Imputation under Green AI Umbrella

This repository contains the codebase for the paper: Missing Data Imputation under Green AI Umbrella presented in Encontro Nacional de Inteligência Artificial e Computacional (ENIAC) 2025

Paper Details

• Authors: Arthur Dantas Mangussi, Ricardo Cardoso Pereira, Pedro Henriques Abreu, and Ana Carolina Lorena
• Abstract: Missing data is a common issue that can undermine machine learning performance, and imputation methods have emerged as state-of-the-art solutions. However, training these methods can be costly and environmentally impactful. In this work, we investigate the missing data problem under Green AI constraints using a Data-Centric AI approach. We evaluate three missingness mechanisms, four missing rates, and ten datasets to assess both data quality and downstream performance. We also propose an optimization model to select the best-performing imputation method while considering sustainability constraints, offering a path toward more responsible and effective data imputation.
• DOI: https://doi.org/10.5753/eniac.2025.14312

Contribuitions

Contributions are welcome! Feel free to open issues, submit pull requests, or provide feedback.

Citation

@inproceedings{eniac,
 author = {Arthur Mangussi and Ricardo Pereira and Pedro Abreu and Ana Lorena},
 title = { Missing Data Under Green AI Umbrella},
 booktitle = {Anais do XXII Encontro Nacional de Inteligência Artificial e Computacional},
 location = {Fortaleza/CE},
 year = {2025},
 keywords = {},
 issn = {2763-9061},
 pages = {1021--1032},
 publisher = {SBC},
 address = {Porto Alegre, RS, Brasil},
 doi = {10.5753/eniac.2025.14312},
 url = {https://sol.sbc.org.br/index.php/eniac/article/view/38788}
}

Acknowledgements

This study was financed, in part, by the Sao Paulo Research Foundation (FAPESP), Brasil. ˜ Process Numbers 2021/06870-3 and 2024/23791-8. This work was also financed through national funds by FCT - Fundac¸ao para a Ci ˜ encia e a Tecnologia, I.P., in the framework ˆ of the Project UIDB/00326/2025 and UIDP/00326/2025. Additionally, it was supported by the Portuguese Recovery and Resilience Plan (PRR) through project C645008882- 00000055-Center for Responsable AI.