In infinite failure non-homogeneous Poisson process model, the hazard function can follow a constant
pattern independent of the failure time, or follow a pattern of non-incremental or non-decreasing. In this
study, we are based on non-homogeneous Poisson Process with infinite number of faults in the process of
removing or correcting software defects. Since the life distribution can use various scale parameters, we use
the extended Weibull distribution and the Weibull extension distribution which are widely used in the
reliability field. Algorithm to estimate the parameters used to maximum likelihood estimator.The result of
this paper, for the mean squared error which measures the difference between the actual value and the
predicted value, because the Weibull extension distribution than extended Weibull distribution and basic
Musa-Okumoto model is the smaller, the Weibull extension distribution is considered to be more efficient.
In addition, for the coefficient of determination representing the explanatory power of the difference in the
predicted value, because the Weibull extension distribution than extended Weibull distribution and basic
Musa-Okumoto model is the higher, the Weibull extension distribution is can be regarded as an efficient
model in terms of the predicted value.