MAHAKIL: Diversity based oversampling approach to alleviate the class imbalance issue in software defect prediction

Kwabena E. Bennin; Jacky Keung; Passakorn Phannachitta; Akito Monden; Solomon Mensah

doi:10.1145/3180155.3182520

MAHAKIL: Diversity based oversampling approach to alleviate the class imbalance issue in software defect prediction

Kwabena E. Bennin, Jacky Keung, Passakorn Phannachitta, Akito Monden, Solomon Mensah

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

5 Citations (Scopus)

Abstract

This study presents MAHAKIL, a novel and efficient synthetic over-sampling approach for software defect datasets that is based on the chromosomal theory of inheritance. Exploiting this theory, MAHAKIL interprets two distinct sub-classes as parents and generates a new instance that inherits different traits from each parent and contributes to the diversity within the data distribution. We extensively compare MAHAKIL with five other sampling approaches using 20 releases of defect datasets from the PROMISE repository and five prediction models. Our experiments indicate that MAHAKIL improves the prediction performance for all the models and achieves better and more significant pf values than the other oversampling approaches, based on robust statistical tests.

Original language	English
Title of host publication	Proceedings of the 40th International Conference on Software Engineering, ICSE 2018
Publisher	IEEE Computer Society
Number of pages	1
Volume	Part F137142
ISBN (Electronic)	9781450356381
DOIs	https://doi.org/10.1145/3180155.3182520
Publication status	Published - May 27 2018
Event	40th International Conference on Software Engineering, ICSE 2018 - Gothenburg, Sweden Duration: May 27 2018 → Jun 3 2018

Other

Other	40th International Conference on Software Engineering, ICSE 2018
Country/Territory	Sweden
City	Gothenburg
Period	5/27/18 → 6/3/18

Keywords

Class imbalance learning
Classification problems
Data sampling methods
Software defect prediction
Synthetic sample generation

ASJC Scopus subject areas

Software

Access to Document

10.1145/3180155.3182520

Cite this

MAHAKIL : Diversity based oversampling approach to alleviate the class imbalance issue in software defect prediction. / Bennin, Kwabena E.; Keung, Jacky; Phannachitta, Passakorn et al.

Proceedings of the 40th International Conference on Software Engineering, ICSE 2018. Vol. Part F137142 IEEE Computer Society, 2018.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Bennin, KE, Keung, J, Phannachitta, P, Monden, A & Mensah, S 2018, MAHAKIL: Diversity based oversampling approach to alleviate the class imbalance issue in software defect prediction. in Proceedings of the 40th International Conference on Software Engineering, ICSE 2018. vol. Part F137142, IEEE Computer Society, 40th International Conference on Software Engineering, ICSE 2018, Gothenburg, Sweden, 5/27/18. https://doi.org/10.1145/3180155.3182520

@inproceedings{780c964bbc9f4ccb968d5349e8c0cd81,

title = "MAHAKIL: Diversity based oversampling approach to alleviate the class imbalance issue in software defect prediction",

abstract = "This study presents MAHAKIL, a novel and efficient synthetic over-sampling approach for software defect datasets that is based on the chromosomal theory of inheritance. Exploiting this theory, MAHAKIL interprets two distinct sub-classes as parents and generates a new instance that inherits different traits from each parent and contributes to the diversity within the data distribution. We extensively compare MAHAKIL with five other sampling approaches using 20 releases of defect datasets from the PROMISE repository and five prediction models. Our experiments indicate that MAHAKIL improves the prediction performance for all the models and achieves better and more significant pf values than the other oversampling approaches, based on robust statistical tests.",

keywords = "Class imbalance learning, Classification problems, Data sampling methods, Software defect prediction, Synthetic sample generation",

author = "Bennin, {Kwabena E.} and Jacky Keung and Passakorn Phannachitta and Akito Monden and Solomon Mensah",

year = "2018",

month = may,

day = "27",

doi = "10.1145/3180155.3182520",

language = "English",

volume = "Part F137142",

booktitle = "Proceedings of the 40th International Conference on Software Engineering, ICSE 2018",

publisher = "IEEE Computer Society",

address = "United States",

note = "40th International Conference on Software Engineering, ICSE 2018 ; Conference date: 27-05-2018 Through 03-06-2018",

}

TY - GEN

T1 - MAHAKIL

T2 - 40th International Conference on Software Engineering, ICSE 2018

AU - Bennin, Kwabena E.

AU - Keung, Jacky

AU - Phannachitta, Passakorn

AU - Monden, Akito

AU - Mensah, Solomon

PY - 2018/5/27

Y1 - 2018/5/27

N2 - This study presents MAHAKIL, a novel and efficient synthetic over-sampling approach for software defect datasets that is based on the chromosomal theory of inheritance. Exploiting this theory, MAHAKIL interprets two distinct sub-classes as parents and generates a new instance that inherits different traits from each parent and contributes to the diversity within the data distribution. We extensively compare MAHAKIL with five other sampling approaches using 20 releases of defect datasets from the PROMISE repository and five prediction models. Our experiments indicate that MAHAKIL improves the prediction performance for all the models and achieves better and more significant pf values than the other oversampling approaches, based on robust statistical tests.

AB - This study presents MAHAKIL, a novel and efficient synthetic over-sampling approach for software defect datasets that is based on the chromosomal theory of inheritance. Exploiting this theory, MAHAKIL interprets two distinct sub-classes as parents and generates a new instance that inherits different traits from each parent and contributes to the diversity within the data distribution. We extensively compare MAHAKIL with five other sampling approaches using 20 releases of defect datasets from the PROMISE repository and five prediction models. Our experiments indicate that MAHAKIL improves the prediction performance for all the models and achieves better and more significant pf values than the other oversampling approaches, based on robust statistical tests.

KW - Class imbalance learning

KW - Classification problems

KW - Data sampling methods

KW - Software defect prediction

KW - Synthetic sample generation

UR - http://www.scopus.com/inward/record.url?scp=85049405348&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85049405348&partnerID=8YFLogxK

U2 - 10.1145/3180155.3182520

DO - 10.1145/3180155.3182520

M3 - Conference contribution

AN - SCOPUS:85049405348

VL - Part F137142

BT - Proceedings of the 40th International Conference on Software Engineering, ICSE 2018

PB - IEEE Computer Society

Y2 - 27 May 2018 through 3 June 2018

ER -

MAHAKIL: Diversity based oversampling approach to alleviate the class imbalance issue in software defect prediction

Abstract

Other

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this