为了能准确评估软件项目的工作量,提高项目按时完工的可能性。本文提出基于定量回归分析的平均生产率法。通过统计历史数据,并运用工具进行统计分析,获得得了软件项目的平均生产率和阶段生产率。分析结果表明软件项目的工作量和规模呈正相关。为了进一步增强平均生产率数值的可用性和可信度,运用数理统计的方法对平均生产率的可信度和置信区间进行求解。通过实验表明平均生产率具有较大的实用性,可以在实际操作中推广使用。可以使用本文所提出的方法对软件项目工作量的进行准确的估计。

To estimate the effort of software projects exactly and improve the compltation rate of software projects, software project average productivity rate based on quantitative calculation was studied.The average software productivity rate and the phase productivity rate were obtained using statistical analysis tools by the collection of historical data. The analysis result indicated that the pertinence between software project effort and size is of positive correlation. To define the reliability and usability of the average productivity rate value, statistics analysis method was used to seek for the reliability and the reliability interval of the average productivity rate. The actual result proved that the average productivity rate obtained is usable, and it could be applied in actual software projects. Therefore the proposed method could be used exactly estimation the effort of software projects.

TP311.5

国家自然科学基金项目（71072123）：基于精益原则的复杂研发项目价值流优化与实证研究；教育部人文社科基金项目（11YJC630384）：基于重大偏差标准的软件项目动态估算模型体系研究