Intrinsic Feature Sensitive Software Reusability Prediction Using Parallelized Feature Selection and Heterogeneous Ensemble Learning Method

The rising global competitiveness in information technology industry has triggered industry to achieve and contribute efficient software solution even at the lower cost. With such a demand, the use of free open source software (FOSS) and component-reusability have become common in software industry. Though, software component reusability can be advantageous; improper reuse might even cause fault, smell, failure and eventual massive losses. Identifying a class with no-reusability, a developer can avoid it to inculcate in program that can improve software reliability. Unlike manual testing methods, which are often criticized for high cost and time-consumption, we propose a novel machine learning based automatic software reusability prediction that predicts each class as REUSABLE or NON-REUSABLE.