A Systematic Approach to Predict System Testing Defects using Prior Phases Metrics for V-Model

Abstract

In order to prevent more defects from escaping to end-users for a V-model development
process, independent testing team needs a predicted total number of defects for any
software under test at the start of system testing so that defects can be fixed as early as
possible. Metrics from requirement to coding phases are required to develop this defect
prediction. Thus, this research introduces and explains the systematic approach to
predict system testing defects for a V-model by using prior phases metrics. By applying
regression analysis as part of the approach, it demonstrates that total number of defects
in system testing can be predicted by using requirement, design and coding metrics. The
approach produces a mathematical equation which is used to predict defects in system
testing. The equation is then verified on new software projects so that it is fit for final
implementation and integration into software development process.