Logistic曲线在软件开发质量预测中的应用研究

摘要

The factors affecting the quality of software are also influenced by other complications apart from the diversity of development modes.For software development projects with multi-phases, constant developing and continual tests, to follow up the overall test quality of the project is of significance to the project quality control.Our study found that the shapes of the time curves of cumulative defect values tested in software development projects are basically in accord with the curves of Logistic and Gompertz function.By adopting VBA programming to traverse three points of all the measured data, three curve parameters ( L, b, a) of the measured data which have the best fitting degrees with two function curves ( the least squares) respectively can be computed.Among them the L value ( saturation value) of Logistic curve can be used to predict the cumulative defect values of the software development project system when it is stable.By analysing the measured cumulative defect values and the prediction values of the functional curves ( determined by three parameters) of the software project under development and after being released and running, it is found that the functional curve can be used to predict and monitor the quality of the software in its development process and after the system has been released.%影响软件质量的因素除了开发方式多种多样外，还受其他因素影响。对于多阶段、不断开发、不断测试的软件开发项目，跟踪项目整体的测试质量对项目的质量控制有重要意义。研究发现软件开发项目中测试出的缺陷累计值的时间曲线基本符合Lo-gistic与Gompertz函数曲线。采用VBA编程，遍历所有实测数据的三点可求解出实测数据分别与两条函数曲线拟合度最好（最小2乘法）的三个曲线参数（ L，b，a）。其中Logistic曲线的L值（即饱和值）可用于预测软件开发项目系统稳定时的缺陷累计值。通过分析软件项目开发中及系统发布运行后的累计缺陷的实测值与函数曲线（三个参数决定的曲线）的预测值，发现该函数曲线可用于预测及监控软件开发过程中及系统发布后的软件质量。