A Prediction Model for System Testing Defects using Regression Analysis
This provides an early quality indicator for software entering system testing, helping testing teams manage activities, but it is incremental as it applies existing regression methods to a specific domain.
The researchers tackled the problem of predicting defects in system testing by developing a regression model using metrics from prior phases, achieving a model with p-value < 0.05 and R-squared > 90%.
This research describes the initial effort of building a prediction model for defects in system testing carried out by an independent testing team. The motivation to have such defect prediction model is to serve as early quality indicator of the software entering system testing and assist the testing team to manage and control test execution activities. Metrics collected from prior phases to system testing are identified and analyzed to determine the potential predictors for building the model. The selected metrics are then put into regression analysis to generate several mathematical equations. Mathematical equation that has p-value of less than 0.05 with R-squared and R-squared (adjusted) more than 90% is selected as the desired prediction model for system testing defects. This model is verified using new projects to confirm that it is fit for actual implementation.