Comparing between Maximum Likelihood and Least Square Estimators for Gompertz Software Reliability Model

Software reliability models (SRMs) are very important for estimating and predicting software reliability in the testing/debugging phase. The contributions of this paper are as follows. First, a historical review of the Gompertz SRM is given. Based on several software failure data, the parameters of the Gompertz software reliability model are estimated using two estimation methods, the traditional maximum likelihood and the least square. The methods of estimation are evaluated using the MSE and R-squared criteria. The results show that the least square estimation is an attractive method in term of predictive performance and can be used when the maximum likelihood method fails to give good prediction results.