著者
カマルゴ クルズ アナ エリカ 落水 浩一郎
出版者
一般社団法人電子情報通信学会
雑誌
電子情報通信学会技術研究報告. SS, ソフトウェアサイエンス (ISSN:09135685)
巻号頁・発行日
vol.107, no.505, pp.43-48, 2008-02-25

Several studies, in the field of object-oriented software quality, have been performed to define models for predicting fault-prone code. However, their predictions take place after the implementation phase, using design-complexity metrics measured from the code. The primary aim of this paper is to provide the foundations for building a model, that predicts fault-prone code in the early phases of the life cycle of the software, using UML metrics. We found that some UML metrics, approximations of traditional design-complexity metrics, can bew acceptable predictors of fault-proneness, as they showed similar performance to the performance of those metrics measured from the implementation.