L公司基于Scrum流程的软件质量研究

发布时间：2018-08-30 10:25

【摘要】：敏捷软件开发流程是一种适应性的软件开发方法,强调快速的迭代,对客户需求变化的及时响应。敏捷开发因其适应了当前软件行业业务形态变化迅速,业界竞争十分激烈等环境因素,逐渐成为业界特别是互联网企业首选的软件开发流程。敏捷开发方法中的Scrum成为主流方法。软件开发中的质量管理是企业非常重视的问题,其中的软件缺陷管理是软件开发日常管理的核心工作。目前针对敏捷流程的缺陷研究,特别是定量的研究还比较少。本文从L公司的项目管理系统中收集了采用Scrum流程开发的两个项目历时两年多的数据,包括缺陷,用户故事,任务等。通过采用零膨胀的负二项回归模型对这些数据进行分析,研究了开发人员在一个迭代中某模块产生的缺陷数量与他所负责的不同类型的任务数量、开发者在上迭代产生的缺陷数量、模块的历史缺陷密度和工作年限等自变量之间的相关关系。通过数据分析得出的结论是；不同类型的开发任务和本迭代的缺陷数量具有不同程度的正相关性,其中业务逻辑层的任务数量和缺陷数量的相关性最强；开发者在上个迭代的缺陷数和模块的历史缺陷密度也与本迭代缺陷数量具有较强的正相关性；工作年限和缺陷数量没有太大相关性。此模型可应用于标识出迭代中可能出现的高缺陷模块供开发和测试人员参考。本文随后结合Scrum开发流程中产生的数据,运用PSM的度量设计思想,通过分析Scrum中不同角色的信息需要,设计了一系列可用于Scrum敏捷开发流程的软件质量度量。
[Abstract]:Agile software development process is an adaptive software development method which emphasizes rapid iteration and timely response to changes in customer requirements. Agile development has gradually become the preferred software development process for the software industry, especially the Internet enterprises, because it adapts to the environment factors such as the rapid changes of the current software industry business forms and the fierce competition in the industry. Scrum in agile development methods has become a mainstream approach. Quality management in software development is a very important problem in enterprises, in which software defect management is the core of daily management of software development. At present, there are few researches on the defects of agile process, especially quantitative research. This paper collects more than two years' data from the project management system of L Company which is developed by Scrum process, including defects, user stories, tasks and so on. By using the zero expansion negative binomial regression model to analyze these data, the number of defects produced by a developer in an iteration and the number of different types of tasks he is responsible for are studied. The correlation between the number of defects generated by the developer in the last iteration, the historical defect density of the module and the working life of the independent variables. Through the data analysis, it is concluded that different types of development tasks and the number of defects in this iteration have different degrees of positive correlation, in which the business logic layer of the number of tasks and the number of defects is the strongest correlation; The number of defects in the last iteration and the historical defect density of the module also have a strong positive correlation with the number of defects in this iteration. This model can be used to identify potentially high-defect modules in the iteration for reference by developers and testers. Combining the data generated in the Scrum development process and using the metric design idea of PSM, this paper designs a series of software quality metrics that can be used in the Scrum agile development process by analyzing the information needs of different roles in Scrum.
【学位授予单位】：上海交通大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP311.5

【相似文献】