I. Introduction
Ever increasing size and complexity of software demands the full proof quality assurance and bug-free product delivery in time. SDP is a vital tool to support the decision making in software development process. It reduces the testing efforts and the cost of development by pointing the poor-quality modules out. It allows to focus testing on these buggy modules and chances to debug them increases. The early detection and correction of defects causes less cost than later stages. Software failure causes huge money loss for an instance the failure of NASA spacecraft caused the loss of $125 million. It was the result of a small data conversion bug [1]. Hence, SDP is essential to enhance the quality of product. The imbalanced state of dataset adversely affects the performance of SDP classifiers. Class imbalance is the situation when the datapoints of one class (negative cases) overwhelms the count of another class (positive cases). The solutions to deal with class-imbalance issue are applied on two levels- 1) data-level and 2) algorithmic level. This work proposes algorithmic level solution to class-imbalance problem using a customized stacking-based ensemble classifier.