Anita Devi
Saurabh Charaya
Abstract
Lots of effort has been used to minimize the software maintenance level of any software specially developed with object-oriented methodology. In this paper, we have proposed a software maintainability prediction model for object-oriented systems using Deep learning. Google Collaboratory has been used with UQES and UIMS datasets for experimental purposes. MAE, MSE, and Varscore have been used as performance metrics. From the results, it was found that varscore for both datasets increases with an increase in the number of hidden layers. However, For the UIMS dataset, the accuracy of the model is 93.7% whereas, for the QUES dataset, the accuracy of the model is 73.12%
Keywords- Software Maintainability, Object-Oriented Systems, Neural Network, Deep Neural Network, MAE and MSE.