Estimating metrics-based software reliability models with elastic net
Yi Li, H. Okamura, Jinke Zheng, Takeyoshi Dohi · 2024
This paper presents a novel framework for Software Reliability Growth Models (SRGMs) that incorporates software metrics. We integrate classical Poisson regression-based fault prediction with Non-Homogeneous Poisson Process (NHPP)-based SRGMs. A key feature of our approach is the simultaneous handling of time-series data for fault detection and software metrics across multiple modules. We outline a modeling framework that combines Poisson regression-based fault prediction with SRGMs and develop an efficient parameter estimation algorithm using the Expectation-Maximization (EM) method.