Detecting Simpson’s Paradox: A Step Towards Fairness in Machine Learning

Rahul Sharma; Minakshi Kaushik; Sijo Arakkal Peious; Markus Bertl; Ankit Vidyarthi; Ashwani Kumar; Dirk Draheim

doi:10.1007/978-3-031-15743-1_7

Detecting Simpson’s Paradox: A Step Towards Fairness in Machine Learning

Rahul Sharma^*, Minakshi Kaushik, Sijo Arakkal Peious, Markus Bertl, Ankit Vidyarthi, Ashwani Kumar, Dirk Draheim

^*Corresponding author for this work

Publication: Chapter in book/Conference proceeding › Contribution to conference proceedings

Abstract

In the last two decades, artificial intelligence (AI) and machine learning (ML) have grown tremendously. However, understanding and assessing the impacts of causality and statistical paradoxes are still some of the critical challenges in their domains. Currently, these terms are widely discussed within the context of explainable AI (XAI) and algorithmic fairness. However, they are still not in the mainstream AI and ML application development scenarios. In this paper, first, we discuss the impact of Simpson’s paradox on linear trends, i.e., on continuous values, and then we demonstrate its effects via three benchmark training datasets used in ML. Next, we provide an algorithm for detecting Simpson’s paradox. The algorithm has experimented with the three datasets and appears beneficial in detecting the cases of Simpson’s paradox in continuous values. In future, the algorithm can be utilized in designing a certain next-generation platform for fairness in ML.

Original language	English
Title of host publication	New Trends in Database and Information Systems - ADBIS 2022 Short Papers, Doctoral Consortium and Workshops
Subtitle of host publication	DOING, K-GALS, MADEISD, MegaData, SWODCH 2022, Proceedings
Editors	Silvia Chiusano, Tania Cerquitelli, Robert Wrembel, Kjetil Nørvåg, Barbara Catania, Genoveva Vargas-Solar, Ester Zumpano
Place of Publication	Cham
Publisher	Springer
Pages	67-76
Number of pages	10
ISBN (Electronic)	9783031157431
ISBN (Print)	9783031157424
DOIs	https://doi.org/10.1007/978-3-031-15743-1_7
Publication status	Published - 2022
Externally published	Yes
Event	3rd Workshop on Intelligent Data - From Data to Knowledge, DOING 2022, 1st Workshop on Knowledge Graphs Analysis on a Large Scale, K-GALS 2022, 4th Workshop on Modern Approaches in Data Engineering and Information System Design, MADEISD 2022, 2nd Workshop on Advanced Data Systems Management, Engineering, and Analytics, MegaData 2022, 2nd Workshop on Semantic Web and Ontology Design for Cultural Heritage, SWODCH 2022 and Doctoral Consortium which accompanied 26th European Conference on Advances in Databases and Information Systems, ADBIS 2022 - Turin, Italy Duration: 5 Sept 2022 → 8 Sept 2022

Publication series

Series	Communications in Computer and Information Science
Volume	1652
ISSN	1865-0929

Conference

Conference	3rd Workshop on Intelligent Data - From Data to Knowledge, DOING 2022, 1st Workshop on Knowledge Graphs Analysis on a Large Scale, K-GALS 2022, 4th Workshop on Modern Approaches in Data Engineering and Information System Design, MADEISD 2022, 2nd Workshop on Advanced Data Systems Management, Engineering, and Analytics, MegaData 2022, 2nd Workshop on Semantic Web and Ontology Design for Cultural Heritage, SWODCH 2022 and Doctoral Consortium which accompanied 26th European Conference on Advances in Databases and Information Systems, ADBIS 2022
Country/Territory	Italy
City	Turin
Period	5/09/22 → 8/09/22

Bibliographical note

Publisher Copyright:
© 2022, Springer Nature Switzerland AG.

Keywords

Artificial intelligence
Big data
Data science
Explainable AI
Machine learning
Simpson’s paradox

Access to Document

10.1007/978-3-031-15743-1_7

Check availability

University Library Catalogue

Cite this

Sharma, R., Kaushik, M., Peious, S. A., Bertl, M., Vidyarthi, A., Kumar, A., & Draheim, D. (2022). Detecting Simpson’s Paradox: A Step Towards Fairness in Machine Learning. In S. Chiusano, T. Cerquitelli, R. Wrembel, K. Nørvåg, B. Catania, G. Vargas-Solar, & E. Zumpano (Eds.), New Trends in Database and Information Systems - ADBIS 2022 Short Papers, Doctoral Consortium and Workshops: DOING, K-GALS, MADEISD, MegaData, SWODCH 2022, Proceedings (pp. 67-76). Springer. https://doi.org/10.1007/978-3-031-15743-1_7

Sharma, Rahul ; Kaushik, Minakshi ; Peious, Sijo Arakkal et al. / Detecting Simpson’s Paradox : A Step Towards Fairness in Machine Learning. New Trends in Database and Information Systems - ADBIS 2022 Short Papers, Doctoral Consortium and Workshops: DOING, K-GALS, MADEISD, MegaData, SWODCH 2022, Proceedings. editor / Silvia Chiusano ; Tania Cerquitelli ; Robert Wrembel ; Kjetil Nørvåg ; Barbara Catania ; Genoveva Vargas-Solar ; Ester Zumpano. Cham : Springer, 2022. pp. 67-76 (Communications in Computer and Information Science, Vol. 1652).

@inproceedings{b8d9dc181d074f8a99529c08e282c521,

title = "Detecting Simpson{\textquoteright}s Paradox: A Step Towards Fairness in Machine Learning",

abstract = "In the last two decades, artificial intelligence (AI) and machine learning (ML) have grown tremendously. However, understanding and assessing the impacts of causality and statistical paradoxes are still some of the critical challenges in their domains. Currently, these terms are widely discussed within the context of explainable AI (XAI) and algorithmic fairness. However, they are still not in the mainstream AI and ML application development scenarios. In this paper, first, we discuss the impact of Simpson{\textquoteright}s paradox on linear trends, i.e., on continuous values, and then we demonstrate its effects via three benchmark training datasets used in ML. Next, we provide an algorithm for detecting Simpson{\textquoteright}s paradox. The algorithm has experimented with the three datasets and appears beneficial in detecting the cases of Simpson{\textquoteright}s paradox in continuous values. In future, the algorithm can be utilized in designing a certain next-generation platform for fairness in ML.",

keywords = "Artificial intelligence, Big data, Data science, Explainable AI, Machine learning, Simpson{\textquoteright}s paradox",

author = "Rahul Sharma and Minakshi Kaushik and Peious, {Sijo Arakkal} and Markus Bertl and Ankit Vidyarthi and Ashwani Kumar and Dirk Draheim",

note = "Publisher Copyright: {\textcopyright} 2022, Springer Nature Switzerland AG.; 3rd Workshop on Intelligent Data - From Data to Knowledge, DOING 2022, 1st Workshop on Knowledge Graphs Analysis on a Large Scale, K-GALS 2022, 4th Workshop on Modern Approaches in Data Engineering and Information System Design, MADEISD 2022, 2nd Workshop on Advanced Data Systems Management, Engineering, and Analytics, MegaData 2022, 2nd Workshop on Semantic Web and Ontology Design for Cultural Heritage, SWODCH 2022 and Doctoral Consortium which accompanied 26th European Conference on Advances in Databases and Information Systems, ADBIS 2022 ; Conference date: 05-09-2022 Through 08-09-2022",

year = "2022",

doi = "10.1007/978-3-031-15743-1_7",

language = "English",

isbn = "9783031157424",

series = "Communications in Computer and Information Science",

publisher = "Springer",

pages = "67--76",

editor = "Silvia Chiusano and Tania Cerquitelli and Robert Wrembel and Kjetil N{\o}rv{\aa}g and Barbara Catania and Genoveva Vargas-Solar and Ester Zumpano",

booktitle = "New Trends in Database and Information Systems - ADBIS 2022 Short Papers, Doctoral Consortium and Workshops",

address = "Germany",

}

Sharma, R, Kaushik, M, Peious, SA, Bertl, M, Vidyarthi, A, Kumar, A & Draheim, D 2022, Detecting Simpson’s Paradox: A Step Towards Fairness in Machine Learning. in S Chiusano, T Cerquitelli, R Wrembel, K Nørvåg, B Catania, G Vargas-Solar & E Zumpano (eds), New Trends in Database and Information Systems - ADBIS 2022 Short Papers, Doctoral Consortium and Workshops: DOING, K-GALS, MADEISD, MegaData, SWODCH 2022, Proceedings. Springer, Cham, Communications in Computer and Information Science, vol. 1652, pp. 67-76, 3rd Workshop on Intelligent Data - From Data to Knowledge, DOING 2022, 1st Workshop on Knowledge Graphs Analysis on a Large Scale, K-GALS 2022, 4th Workshop on Modern Approaches in Data Engineering and Information System Design, MADEISD 2022, 2nd Workshop on Advanced Data Systems Management, Engineering, and Analytics, MegaData 2022, 2nd Workshop on Semantic Web and Ontology Design for Cultural Heritage, SWODCH 2022 and Doctoral Consortium which accompanied 26th European Conference on Advances in Databases and Information Systems, ADBIS 2022, Turin, Italy, 5/09/22. https://doi.org/10.1007/978-3-031-15743-1_7

Detecting Simpson’s Paradox: A Step Towards Fairness in Machine Learning. / Sharma, Rahul; Kaushik, Minakshi; Peious, Sijo Arakkal et al.
New Trends in Database and Information Systems - ADBIS 2022 Short Papers, Doctoral Consortium and Workshops: DOING, K-GALS, MADEISD, MegaData, SWODCH 2022, Proceedings. ed. / Silvia Chiusano; Tania Cerquitelli; Robert Wrembel; Kjetil Nørvåg; Barbara Catania; Genoveva Vargas-Solar; Ester Zumpano. Cham: Springer, 2022. p. 67-76 (Communications in Computer and Information Science, Vol. 1652).

Publication: Chapter in book/Conference proceeding › Contribution to conference proceedings

TY - GEN

T1 - Detecting Simpson’s Paradox

T2 - 3rd Workshop on Intelligent Data - From Data to Knowledge, DOING 2022, 1st Workshop on Knowledge Graphs Analysis on a Large Scale, K-GALS 2022, 4th Workshop on Modern Approaches in Data Engineering and Information System Design, MADEISD 2022, 2nd Workshop on Advanced Data Systems Management, Engineering, and Analytics, MegaData 2022, 2nd Workshop on Semantic Web and Ontology Design for Cultural Heritage, SWODCH 2022 and Doctoral Consortium which accompanied 26th European Conference on Advances in Databases and Information Systems, ADBIS 2022

AU - Sharma, Rahul

AU - Kaushik, Minakshi

AU - Peious, Sijo Arakkal

AU - Bertl, Markus

AU - Vidyarthi, Ankit

AU - Kumar, Ashwani

AU - Draheim, Dirk

PY - 2022

Y1 - 2022

N2 - In the last two decades, artificial intelligence (AI) and machine learning (ML) have grown tremendously. However, understanding and assessing the impacts of causality and statistical paradoxes are still some of the critical challenges in their domains. Currently, these terms are widely discussed within the context of explainable AI (XAI) and algorithmic fairness. However, they are still not in the mainstream AI and ML application development scenarios. In this paper, first, we discuss the impact of Simpson’s paradox on linear trends, i.e., on continuous values, and then we demonstrate its effects via three benchmark training datasets used in ML. Next, we provide an algorithm for detecting Simpson’s paradox. The algorithm has experimented with the three datasets and appears beneficial in detecting the cases of Simpson’s paradox in continuous values. In future, the algorithm can be utilized in designing a certain next-generation platform for fairness in ML.

AB - In the last two decades, artificial intelligence (AI) and machine learning (ML) have grown tremendously. However, understanding and assessing the impacts of causality and statistical paradoxes are still some of the critical challenges in their domains. Currently, these terms are widely discussed within the context of explainable AI (XAI) and algorithmic fairness. However, they are still not in the mainstream AI and ML application development scenarios. In this paper, first, we discuss the impact of Simpson’s paradox on linear trends, i.e., on continuous values, and then we demonstrate its effects via three benchmark training datasets used in ML. Next, we provide an algorithm for detecting Simpson’s paradox. The algorithm has experimented with the three datasets and appears beneficial in detecting the cases of Simpson’s paradox in continuous values. In future, the algorithm can be utilized in designing a certain next-generation platform for fairness in ML.

KW - Artificial intelligence

KW - Big data

KW - Data science

KW - Explainable AI

KW - Machine learning

KW - Simpson’s paradox

UR - http://www.scopus.com/inward/record.url?scp=85138005682&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-15743-1_7

DO - 10.1007/978-3-031-15743-1_7

M3 - Contribution to conference proceedings

AN - SCOPUS:85138005682

SN - 9783031157424

T3 - Communications in Computer and Information Science

SP - 67

EP - 76

BT - New Trends in Database and Information Systems - ADBIS 2022 Short Papers, Doctoral Consortium and Workshops

A2 - Chiusano, Silvia

A2 - Cerquitelli, Tania

A2 - Wrembel, Robert

A2 - Nørvåg, Kjetil

A2 - Catania, Barbara

A2 - Vargas-Solar, Genoveva

A2 - Zumpano, Ester

PB - Springer

CY - Cham

Y2 - 5 September 2022 through 8 September 2022

ER -

Sharma R, Kaushik M, Peious SA, Bertl M, Vidyarthi A, Kumar A et al. Detecting Simpson’s Paradox: A Step Towards Fairness in Machine Learning. In Chiusano S, Cerquitelli T, Wrembel R, Nørvåg K, Catania B, Vargas-Solar G, Zumpano E, editors, New Trends in Database and Information Systems - ADBIS 2022 Short Papers, Doctoral Consortium and Workshops: DOING, K-GALS, MADEISD, MegaData, SWODCH 2022, Proceedings. Cham: Springer. 2022. p. 67-76. (Communications in Computer and Information Science, Vol. 1652). doi: 10.1007/978-3-031-15743-1_7