Classification of NFR based Importance Level of User Story in Agile Software Development

The fact that requirement engineering creates effective technology and prevents system failure is very important in the early phases of software development. The final implementation criteria and the selection of many alternative models both heavily depend on non-functional requirements. The term "non-functional requirements" (NFR) refers to a group of qualities that software must possess, such as security, dependability, performance, etc. Extracting and considering NFR from user stories in Agile Software Development (ASD) can help developers deliver quality software that meets the expectations of users completely. This paper proposed the classification of NFR based importance level of user story in ASD using machine learning models. At the beginning of this work, requirement converted into user stories, then the weight of NFR type and importance level of each user story calculated and classified using machine learning models. The outcome of this work shows that random forest classifier and XGBoost are achieved the best results with an accuracy of 96.39% and decision tree classifier accuracy is 96.13%, with comparisons of K-nearest neighbor, linear discriminant analysis, support vector classifier, logistic regression, and naïve bayes.