Sensitivity Analysis in Optimal Software Release Time Problems

M. Xie,B. Yang,O. Gaudoin

DOI: https://doi.org/10.1007/bf03398848

2004-01-01

Abstract:In the modeling of software reliability and in the decision-making using software reliability models, there are a number of parameters whose values are either assumed or estimated based on early data. These parameters are usually inaccurate and thus it is important to study the sensitivity of the estimation. In this paper, the parameter sensitivity problem is studied from two different points of view. First, in addition to point estimation that is commonly used in data analysis, interval estimation is recommended to quantify the errors in the estimated parameters. Second, with the example of sensitivity analysis for the release time determination problem that usually involves both software reliability model parameters and software cost model parameters, a statistical design of experiments technique is employed. This method can not only estimate the effect of separate change of each model parameter, but also estimate the interaction effects when parameters are changed simultaneously. It is useful information for decision maker so that important parameters can be estimated more accurately.