Sébastien Rufiange

A common problem in software engineering is to find the right balance between quality and cost, while delivering on time. Needs and technology are evolving faster than our skill to manage such projects. To manage and implement efficiently these change requests can affect the software quality and maintenance costs.

In this study, we used several data mining techniques (classification and regression) to find a relationship between change impact (e.g. lines of code, number of modified files) and software design measures. We found out which attributes seems to have more impact on the maintenance effort. The data was collected from open-source projects, but a larger scale study is currently being done.

References:

• S. Rufiange. "Étude de l’impact des changements sur des logiciels libres à l’aide de techniques de forage de données". 77e congrès de l'ACFAS. Ottawa. 2009.