Experience-based project management in a distributed development environment.

摘要

Experience-based project management utilizes measurement data from past and ongoing projects to characterize, evaluate and improve software processes and their products. In an environment where common resources are distributed over various physical locations (such as world-wide software development or subcontract management), it is often difficult to obtain and utilize the data required to support project management. This dissertation work focuses on two main topics: building baseline models from highly variable time-series data; and defining contexts for the data to ensure consistency when combining the data in a distributed environment. The concepts and techniques that have been developed to support experience-based project management in a distributed environment are implemented in a Web-based system called the Web Measurement Environment (WebME). For building baseline models, WebME uses a technique based on a stock market indicator to identify trends and describe the behavior of attribute data from a single project. This technique produces a “characteristic curve” of an attribute (e.g., effort, number of defects, size) over the lifecycle of a single project. Projects with similar characteristic curves are combined to build baseline models of expected behavior. In addition to this analysis technique, the structural model of measurement used by WebME allows the user to specify context properties of the data such as location, access methods, and the units and intervals of collection.; The WebME tool and the techniques have been applied to data sets from five different environments, the Software Engineering Laboratory (SEL) at NASA's Goddard Space Flight Center, the Rapid Development Laboratory (RDL) at NASA's Johnson Space Center, a large consulting company, a large telecommunications company and a graduate level software engineering course taught at the University of Maryland. The results from these data sets demonstrate how these techniques and the tool can be used to specify the context of the data and uncover interesting information from the data to support project management.