NFR Elicitation and Priority Prediction in ASD

Requirement engineering is crucial in the early stages of software development because it produces useful technology and guards against system failure. Non-functional requirements have a major role in both the selection of several different models and the final implementation criteria. The phrase “non-functional requirements” (NFRs) describes a set of characteristics, such security, reliability, performance, etc., that software has to have. When using Agile Software Development (ASD), developers may produce high-quality software that fully satisfies customer expectations by extracting and taking into account NFR from user stories. This paper proposed the NFR elicitation, prioritization process and prediction for the user stories. In initial phase of elicitation process, the baseline NFR counts and added new keywords NFR counts are calculated on the behalf of standard keyword and online library from the user stories with involvement of experts and users. The keyword density weight and importance value are calculated on behalf of occurrences of NFRs in second phase. Through importance value also set the priority level low, medium and high to the user stories. In final phase, the classification models are implemented on priority level of user stories. The result and outcome of NFR elicitation is improved by 6.74% and outcome of priority prediction is done by various classification models where the results shows that XGBoost gives 95.16% and random forest classifier gives 94.95% accuracy.