Rough set-based data analysis in goal-oriented software measurement

The analysis of software engineering data is often concerned with the treatment of incomplete knowledge, the management of inconsistent pieces of information and the manipulation of various data representation levels. Existing techniques of data analysis are mainly based on quite strong assumptions (some knowledge about dependencies, probability distributions, and a large number of experiments), are unable to derive conclusions from incomplete knowledge, or cannot manage inconsistent pieces of information. A rough set is a collection of objects which, in general, cannot be precisely characterized in terms of the values of the set of attributes, while a lower and an upper approximation of the collection can do so. Rough sets have been successfully applied for data analysis in different areas. In this paper, the approach is applied to the analysis of software engineering data resulting from goal-oriented measurement. Fundamental principles and concepts of rough sets are presented. They are illustrated by an example predicting the criticality of software modules based on metrics data from the early development phases. In a further application, analysis of COCOMO (COnstructive COst MOdel) cost drivers is studied.