Adopting Six Sigma approach in predicting functional defects for system testing

This research focuses on constructing a mathematical model to predict functional defects in system testing by applying Six Sigma approach. The motivation behind this effort is to achieve zero known post release defects of the software delivered to end-user. Besides serving as the indicator of optimizing testing process, predicting functional defects at the start of testing allows testing team to put comprehensive test coverage, find as many defects as possible and determine when to stop testing so that all known defects are contained within testing phase. Design for Six Sigma (DfSS) is chosen as the methodology as it emphasizes on customers' requirement and systematic techniques to build the model. Historical data becomes the crucial elements in this study. Metrics related to potential predictors and their relationships for the model are identified, which focuses on metrics in phases prior to testing phase. Repeatability and capability of testers' consistency in finding defects are analyzed. Type of data required are also identified and collected. The metrics of selected predictors which incorporate testing and development metrics are measured against total functional defects using multiple regression analysis. The best and most significant mathematical model generated by the regression analysis is selected as the proposed prediction model for functional defects in system testing phase. Validation of the model is then conducted to prove the goodness for implementation. Recommendation and future research work are provided at the end of this study.