Software bug identification and prediction through software metrics in object oriented projects an empirical analysis

Abstract

In the software engineering, quality assurance plays an important role. newlineThe quality assurance as an activity, observes the execution of software project to ensure that the behavior of product is in accordance with the expectations. The testing is associated with quality assurance activities. The testing takes a lot of time and an effort of the tester to test the test newlinecases. Even after enough manual or automatic testing, bugs remain uncovered because of lack of time. So, a need arises to focus on this area to save the time and cost of the organizations. The software developer or newlinetester should be aware about the main reasons of software bugs so that they can focus on the right part of the code at the right time. Need of introducing product, process and project metrics is also very essential for newlinethe identification of major causes of bugs. Predictions will always be best if the history of project is taken into consideration. We can come up with accurate predictors with the help of root causes of the software bugs. Several bug prediction models can use bug indicators as the input of model to predict the number of bugs. newlinePrediction attempts to provide quantitative measures to help the software testers and developers. With more number of bug indicators, a step can be taken towards wider horizon of bug prediction thus enabling higher devotion to improve quality of software products. Therefore, identification of several reasons of software bugs and implementation of effective bug prediction models are needed to widen the scope of bug newlineprediction approaches and to improve the software quality. After estimating the future bugs using prediction models, awareness of bug severity is also required to avoid the expected harms to software products. newlineIntroduction of Artificial Neural Network (ANN) was needed to improve the prediction potential. In this work an attempt has been made to associate different levels and types of inheritance through neural network newlineby establishing a correlation framework with diverse types of bug severitie.