Bootstrap Interval Estimation Methods for Cost-Optimal Software Release Planning
Shinji Inoue, Shigeru Yamada · 2013
We discuss interval estimation methods for cost-optimal software release time based on a discretized software reliability growth model. In our approach, we use a bootstrap method, in which we do not need to derive probability distributions of model parameters and optimal software release time analytically by using an asymptotic theory assuming a large number of samples. Then we estimate bootstrap confidence intervals of cost-optimal software release time based on two kinds of bootstrap confidence interval methods. Our numerical examples confirm that our bootstrap approach yields a simulation-based probability distribution of cost-optimal software release time from software fault-count data.