软件可靠性预测的ARIMA方法研究

Forecasting method of software reliability based on ARIMA model

对基于求和自回归滑动平均模型(ARIMA模型)的软件可靠性预测方法进行了研究,提出了将软件可靠性失效数据看作时间序列,通过建立相应的ARIMA(p,d,q)模型来进行预测的方法。对该方法的基本思想、模型表述、建模流程进行了详细介绍,并依据上述方法选用Musa经典数据集中的ProjectSS2中的数据进行了预测,结果表明预测的准确性较高,说明该方法适用于软件可靠性预测。

A method of software reliability forecasting based on the ARIMA （Autoregressive Integrated Moving Average） Model is studied in this paper.It proposes a method which supposes that software failure data can be regarded as time series,and then be forecasted by building correlative ARIMA（p ,d,q ） model.The basic idea,model formulation and modeling procedures are introduced in detail.The data of Project SS2 in Musa classical data set is forecasted by using this method,and the results shows that the forecasting precision is much higher.So the method is suitable for software reliability forecasting.