提出了一种基于数据包络分析的软件任务性能基准评价新方法——TaskBeD.介绍了TaskBeD的任务基准评价模型和核心算法(挖掘高性能的软件任务,建立参考任务集和结果的敏感度分析).实验结果显示,TaskBeD能够高效处理多变元和可变规模收益任务数据.
In this paper, a software task performance benchmarking method based on Data Envelopment Analysis (DEA)-TaskBeD is proposed. TaskBeD's fundamental benchmarking model and algorithms (transforming undesirable outputs, identifying high performance tasks, establishing reference sets and performing sensitivity analysis) are introduced. Experimental results show that the proposed TaskBeD method achieves a good result of dealing with multivariate and VRS.