Abstract

Software maintenance is the most time consuming activity in the life cycle of software. Software maintenance suffers from missed deadlines and from being over budget. Managers usually pay more attention to development than to maintenance: for example, they prefer to assign senior developers to the development phase tasks and neglect maintenance ones. Managers have difficulty identifying problems, and their causes, in maintenance.
This thesis presents techniques for analysis on the proposed release history database to provide metrics for improvement of the maintenance phase. The proposed release history database is enriched by valuable data that comes from an issue tracking system, code repository, and time entry system. The proposed release history database and the analysis of the data contained there provides metrics which allow maintainers to find risky and time consuming codes, recommending maintenance team and maintenance location and a suggestions for the future of the maintenance. Automation is also provided as a proof of concept through a prototypical tool.