针对软件维护的工作量估算,根据统计学原理,给出估算的模型和模型的抽取方法.首先介绍通过变量相关性分析,进行变量的筛选方法,然后提出“分步递进”和“分步递减”模型提取算法.该算法充分融合了专家经验和数学工具.在对模型的评价上,除传统的误差均值、误差方差的指标外,还介绍了通过误差分布情况对模型进行分析和改进的方法.最后通过一个具体实例说明了模型提取全过程.%Aiming at estimation of software maintenance workload, the estimation model and the model extraction method are presented based on the statistics principle. In the paper we first introduce the approach for screening the variants by pertinence analysis of the variants, and then propose the model extraction algorithm of step-by-step progression and step-by-step decline, which fully integrates the expert experiences and the mathematical tools. For the evaluation of the model, apart from the indexes of conventional mean error and error variance, we also introduce the method of model analyses and improvement according to error distribution situation. In end of the paper, a specific example is presented to illustrate the whole process of model extraction.
展开▼
