Bachelor and Master Theses

Title: A New Perspective on Predicting Maintenance Costs
Subject: Computer Science
Level: Advanced
Description: Whether or not a code base should be refactored is a difficult
question. There are always features to be implemented first, and it
never seems to be the right moment. In this thesis, I collect data
from real-life projects that project managers can use to quantify the
benefits of a refactoring, and help them in making a decision.

To get to this data, I measure the the correlation between various
published source code metrics and maintenance cost for software files in 3 Apache open source projects.

The maintenance effort is measured per release, and it is measured in
four ways: the amount of lines of code that had to be changed to
resolve tasks in a release, the amount of discussion that tasks
generated, the amount of patches and commits that were submitted to
resolve a task, and the bug rate of the project during the release.
These 4 effort measures together give us a good impression of the
maintenance effort of the project during a given release.

I then compare a set of generally well published source code metrics
against these maintenance measures to see how well they are able to
predict increases or decreases in effort.
Company: University of Hawaii at Manoa, kontaktperson: Patricia Lago Vu Amsterdam, Rick Kazman Hawaii Univerersity
Prel. end date: 31-04-2012
Presentation date: 07-05-2012
Student: Florian Uunk flori@nuunk.nl
IDT supervisor: Ivica Crnkovic
ivica.crnkovic@mdh.se, +46-21-103183

Rapport och bilagor

Size

Senaste uppdatering

TR1373.pdf

453197

2012-05-08, 00:08


  • Mälardalen University |
  • Box 883 |
  • 721 23 Västerås/Eskilstuna |
  • 021-101300, 016-153600 |
  • webmaster |
  • Latest update: 2017.08.23