Supervised machine learning approach to predict qualitative software product

Software development process (SDP) is a framework imposed on software product development and is a multi-stage process wherein a wide range of tasks and activities pan out in each stage. Each stage requires careful observations to improve productivity, quality, etc. to ease the process of development. During each stage, problems surface likes constraint of on-time completion, proper utilization of available resources and appropriate traceability of work progress, etc. and may lead to reiteration due to the defects spotted during testing and then, results into the negative walk-through due to unsatisfactory outcomes. Working on such defects can help to take a step towards the proper steering of activities and thus to improve the expected performance of the software product. Handpicking the proper notable features of SDP and then analyzing their nature towards the outcome can greatly help in getting a reliable software product by meeting the expected objectives. This paper proposed supervised Machine Learning (ML) models for the predictions of better SDP, particularly focusing on cost estimation, defect prediction, and reusability. The experimental studies were conducted on the primary data, and the evaluation reveals the model suitability in terms of efficiency and effectiveness for SDP prediction (accuracy of cost estimation: 65%, defect prediction: 93% and reusability: 82%).