Statistics Department, Faculty of Sciences, King Abdulaziz University, Jeddah, Kingdom of Saudi Arabia

---

Abstract—The most important attribute of software quality is software reliability. Over the past years, many reliability models have been constructed for measuring and improving the growth of reliability. In reliability engineering, less attention has been paid to find confidence intervals around the estimated parameters and measures of reliability although its importance at producing precise prediction results. In this paper, we focus on the construction of confidence intervals based on a reliability model that belongs to the family of Non-Homogeneous Poisson Process (NHPP) which is created using the Half-Logistic Distribution (HLD). In more detail, Asymptotic Confidence Intervals (ACIs) are derived using the Fisher information matrix for the constructed model’s parameters as well as its intensity and number of remaining errors functions. Analysis of the DACS software reliability dataset to examine the accuracy of the point and interval estimation are provided.

Index Terms—Non-homogeneous Poisson process, half-logistic distribution, intensity function, number of remaining errors, asymptotic confidence interval, Fisher information matrix.

[PDF]

Cite: Lutfiah Ismail Al turk, "Confidence Intervals for the Failure Intensity and Number of Remaining Errors Functions Based on Non-Homogeneous Poisson Process Model," Journal of Software vol. 13, no. 12, pp. 675-686, 2018.