FEPP: IMPROVING SOFTWARE REQUIREMENTS RISK PREDICTION THROUGH HYBRID RULE EXTRACTION AND MULTI-CLASS LEARNING

Predicting risks in software requirements, acritical and vital component of the SoftwareDevelopment Life Cycle (SDLC), is challengingdue to the growing complexity of softwareprojects. An inability to properly predict suchrisks may result in the failure of a softwareproject. Risk prediction is more important insoftware requirements as it is the first phase ofany software project. Therefore, this studyproposes a unique approach for risk prediction insoftware requirements called ForExPlusPlus(FEPP). The proposed model is benchmarkedagainst standard models including K-nearestNeighbour (KNN), Naïve Bayes (NB), LogisticModel Tree (LMT), Random Forest (RF), andSupport Vector Machine (SVM). These modelsare trained using the Zenodo repository dataset,and the outcomes are assessed using commonevaluation standards. The precision, F-measure(FM), and Mathew's correlation coefficient(MCC) are used to critically evaluate themodels' accuracy analysis. The Kappa Statistic(KS) and Mean Absolute Error (MAE) are usedto evaluate the error rate. With an accuracy of96.84%, the recommended FEPP outperformsKNN, which has the lowest accuracy of 50.99%.