Research On Software Quality In Scrum Of L Company

Agile software development method emphasizes adaptability, short iteration and quick response to requirement changes. It becomes more and more popular in software industry because it is suitable for today’s high competitive market and fast changing environment. Scrum is one of the most used agile development method in industry. Quality management is a very important aspect in software development management. Software defect management is the core part in the quality management. There isn’t much quantitative research on defects in agile development.In this thesis we have collected two projects’scrum data from L Company’s project management system. Then we use zero inflated negative binomial regression model to analyze the relationship between developer’s defect count per module in current iteration and other independent variables such as the task count of different type of tasks, historical module defect density, developer’s defect count in previous iteration and developer’s working years.We conclude that the task count of business logic layer has the strongest positive correlation with defect count in current iteration. The working years of developer have very weak correlation with defect count. Other independent variables have different strength of correlation with defect count. This model could be used to identify modules which have potential high defect count.In the second part of this thesis, we analyzed the information needs of different scrum roles, then we constructed several quality metrics for the scrum process using the Practical Software Measurement method.