提高软件成本估计精度的方法

提出了一种估计软件成本的新方法，该方法允许使用一个替代变量去代替变量事先无法知道的软件规模的准确值，并在替代变量已知的情况下，给出了软件成本的估计和置信区间。实践表明，该方法的成本估计值比传统的标准回归分析方法和ＣＯＣＯＭＯ方法更为精确。

Approach to Improve the Calibration of Software Cost Prediction

A new approach to predict the cost of software is presented in this paper. It allows to use a surrogate variable to substitute the actual size of the software project which is unknown at the begining of the project. The estimation of the cost and its confidence interval are also given under the condition that the surrogate variable is known. The practice shows that this approach can obtain more accurate prediction of project cost than traditional standard regression analysis and famous COCOMO method.