During testing of software, most of the bugs lying dormant in the software gets uncovered once the test cases are executed. Different bugs may take different amounts of effort and expertise for their removal. To understand the complexity of bugs from a developer‟s perspective, researchers have developed different mathematical models. Software consists of two types of bugs, dependent and independent. Dependent bugs are those whose removal depends upon the removal of some other bugs on which it is dependent. Dependency of bugs also makes the bug complex and bugs will take more time during fixing. Different debugging time lags functions have been taken to model different complexity of bugs. The aim of this paper is to study the bugs of different complexity. The complexity of bugs has been also modeled using dependency concept. Testing effort dependent bug complexity model using fault dependency has been also discussed. We also feel that that more complex bug will take more time and less complex bug will take less time during fixing. During removal of bugs, the removal team gets more familiar with the code during the fixing. The learning effect during testing has been incorporated using logistic removal rate. The models are validated based on different comparison criteria namely MSE, R2 , Bias, Variation and Root mean squared error.
Keywords/Index Terms: Non-homogeneous Poisson process, bug complexity, bugs types.