Predictive Classification Framework for Software Demand Using Ensembled Machine Learning

Software demands are the primary set of essential textual description in baseline of software engineering associated with design anticipation of clients or stakeholder which are quite challenging to be properly evaluated in case of large quantity of information. Review of existing methodologies exhibits frequent adoption of predictive approaches using manifold learning-based approaches. Hence, this manuscript presents a novel and simplified predictive methodology where ensembled machine learning is used for classifying the novel classes of software demands. The study model uses a simplified transformation approach where the input dataset of software demands is subjected to text mining and feature extraction followed by testifying the extracted features on ensembled machine learning models. The study outcome shows Random Forest to offer higher accuracy, lower processing time, and reduced memory utilization in contrast to Support Vector Machine, Decision Tree, K-Nearest Neighbour, and Naive Bayes.