TY - GEN
T1 - Classification of Testable and Valuable User Stories by using Supervised Machine Learning Classifiers
AU - Subedi, Ishan Mani
AU - Singh, Maninder
AU - Ramasamy, Vijayalakshmi
AU - Walia, Gursimran Singh
N1 - Publisher Copyright:
© 2021 IEEE.
PY - 2021
Y1 - 2021
N2 - Agile is one of the most widely used software development methodologies that include user stories, the smallest units semi-structured specifications to capture the requirements from a user's point of view. Despite being popular, only a little research has been done to automate the quality checking/analysis of a user story before assigning it to a sprint. In this study, we have chosen two metrics, i.e., Testable and Valuable criteria from INVEST checklist, and have applied supervised machine learning classifiers to automatically classify them. Since the industrial data collected for the research was unbalanced, we also applied data balancing techniques such as SMOTE, RUS, ROS, and Back translation (BT) to verify if they improved any classification metrics. Although we did not see any significant improvements in accuracy and precision for the classifiers after applying data balancing techniques, we noticed a significant improvement in recall values across all the classifiers. Our research provides some promising insights into how this research could be used in the software industry to automate the analysis of user stories and improve the quality of software produced.
AB - Agile is one of the most widely used software development methodologies that include user stories, the smallest units semi-structured specifications to capture the requirements from a user's point of view. Despite being popular, only a little research has been done to automate the quality checking/analysis of a user story before assigning it to a sprint. In this study, we have chosen two metrics, i.e., Testable and Valuable criteria from INVEST checklist, and have applied supervised machine learning classifiers to automatically classify them. Since the industrial data collected for the research was unbalanced, we also applied data balancing techniques such as SMOTE, RUS, ROS, and Back translation (BT) to verify if they improved any classification metrics. Although we did not see any significant improvements in accuracy and precision for the classifiers after applying data balancing techniques, we noticed a significant improvement in recall values across all the classifiers. Our research provides some promising insights into how this research could be used in the software industry to automate the analysis of user stories and improve the quality of software produced.
KW - Machine learning
KW - Requirement Engineering and Quality
KW - Text Augmentation
KW - User Stories
UR - http://www.scopus.com/inward/record.url?scp=85126956890&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85126956890&partnerID=8YFLogxK
U2 - 10.1109/ISSREW53611.2021.00111
DO - 10.1109/ISSREW53611.2021.00111
M3 - Conference contribution
AN - SCOPUS:85126956890
T3 - Proceedings - 2021 IEEE International Symposium on Software Reliability Engineering Workshops, ISSREW 2021
SP - 409
EP - 414
BT - Proceedings - 2021 IEEE International Symposium on Software Reliability Engineering Workshops, ISSREW 2021
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 32nd IEEE International Symposium on Software Reliability Engineering Workshops, ISSREW 2021
Y2 - 25 October 2021 through 28 October 2021
ER -