Evolutionary Computing Assisted Heterogenous Ensemble Model for Web-of-Service Software Reusability Prediction

Software being one of the inevitable needs of industries today requires ensuring reliability of computation. On the other hand, to enhance cost of development and productivity software developers use Component based Software Development (CSD) and Free Open Source Software (FOSS) element. Existing functions or software components are used frequently to perform the task. On contrary, the uncontrolled, excessive or improper use of such components lead software faults, aging or smells, which makes overall software vulnerable. To alleviate it, software reusability estimation can be a potential solution. The classical machine learning models can be confined to perform optimal reusability prediction with highly complicate and large size software design such as Web-of-Service (WoS) software. In this paper evolutionary computing assisted ensemble model is percolated that significantly mitigates the key issues of regional minima and convergence. Being OOP software reusability prediction model, six software metrics have been extracted from WoS software, which have been processed with different base learning models. Majority voting scheme in conjunction with Genetic Algorithm based ensemble decision enables the proposed model to exhibit better performance than any other base classifiers.