Abstract Software effort estimation is essential to efficiently managing risks, resource allocation, and project planning in software projects. Despite the widespread usage of traditional estimating approaches, their accuracy frequently needs to be improved due to software development's intricate and dynamic nature. Deep learning methods are now being investigated for software effort estimation due to their outstanding promise in many fields. However, data heterogeneity and noise can constrain the prediction performance of a single deep-learning model. We thoroughly investigate the use of deep learning stacking algorithms for software effort estimation in this research. Stacking, an ensemble learning strategy, uses the combined predictive strength of various base models to balance out individual flaws and improve accuracy overall. Our study focuses on this method's ability to handle problems with software effort estimates and its potential to deliver cutting-edge predictive performance. We assess how well individual deep-learning models perform compared to stacked ensembles and conventional estimation methods. This research thoroughly examines stacking deep learning algorithms for software effort estimates, highlighting their effectiveness in enhancing forecast accuracy and resilience. The conclusions have essential ramifications for managing software projects, enabling better resource allocation, risk avoidance, and more fruitful software development endeavors.
Alan : Mühendislik
Dergi Türü : Uluslararası
Benzer Makaleler | Yazar | # |
---|
Makale | Yazar | # |
---|