Data In Brief _Code Smells_Metrics_Fault_ Data_ECLIPSE

For the development of prediction model the association between metrics, code smells and faulty classes in post release object oriented open source systems (Eclipse IDE) is examined. An array of metrics is used as independent variables which includes diverse characteristics of design. Two type of cataloging for code smells - Class Level and Method Level are performed with a set of eight code smells as dependent variables. Reverse engineering code smell predictor application - iPlasma was used which was able to detect the selected code smells and Object Oriented metrics. The perceptiveness of the model on the whole is considered to be of fair to good quality and the model qualifies to be called as a successful model which may require further performance tuning in terms of data and algorithm parameters.为构建预测模型，本研究针对发布后面向对象开源系统（Eclipse IDE）中的软件度量（Metrics）、代码异味（Code Smell）与缺陷类之间的关联展开分析。本研究选取涵盖多样化设计特征的多组软件度量作为自变量；针对代码异味，本研究采用类级与方法级两类分类方案，并将8种代码异味设定为因变量。本研究采用逆向工程代码异味预测工具iPlasma，该工具可实现选定代码异味与面向对象软件度量的检测。整体而言，该模型的预测性能被评定为中等至优良水平，可被认定为成功的预测模型，但仍需针对数据与算法参数开展进一步的性能调优工作。