A Novel Alextreenet Based Software Defects Prediction with the Human Interacted Gui

Producing high-quality software at an affordable price is always a priority for developers. However, due to exponential growth in size and complexity, it is almost impossible to develop high-quality software at a lower cost. This problem can be solved by identifying defects at an early stage of the development cycle. Recent developments in software defect prediction (SDP) include the integration of screens to create a collaborative approach. The proposed work is the Multi-Scale fusion based AlexTreeNet which is highly efficient framework for feature learning and classification. This paper proposes an innovative AlexTreeNet-based approach integrated with human-assisted Graphical User Interface (GUI) testing for the effective prediction of software defects. Human-assisted GUI testing is the most important part of this work, as it allows testers to intuitively load datasets, visualize defect predictions, and validate results. This interface bridges the gap between automated predictions and human intuition, ensuring comprehensive defect analysis. The proposed system evaluated using NASA and PROMISE datasets is implemented in python. For NASA dataset, the proposed method achieved the highest sensitivity of 0.995 for MC1 dataset, perfect specificity for KC3 dataset as 1.000 and an accuracy of 0.991 for MC1, surpassing the performance of competing models. For PROMISE dataset, AlexTreeNet maintained its superiority, providing the highest sensitivity value of 0.997 for KC1 and an accuracy value of 0.994 for JM1 and KC1 datasets. Thus, this work provides a scalable and accurate early defect detection through the integration of human insights into deep learning methods, which diminishes maintenance cost and enhances reliability in software.