A Study on Comparative Evaluation of Software Reliability Model Applying Modified Exponential Distribution
Hee-Cheul Kim · International Journal of Engineering Research and Technology · 2020
This paper, the reliability of software of lifetime distributions based on non-Homogeneous Poisson process were discussed.The lifetime distributions was used the exponential distribution widely used in the field of software reliability and the inverse-exponential distribution which can be commonly applied in the ecological field , Burr-Hatke-exponential distribution which following decrease hazard function..In the present study, the mean square error of the exponential distribution model is smaller than that of the inverseexponential distribution model and Burr-Hatke-exponential distribution.Also, in the mean function pattern comparison, the exponential distribution model can be seen that the width of the prediction difference is the smallest in the comparison with the true value.Therefore, the exponential distribution model can be regarded as an efficient model in terms of accuracy.However, the estimated value of determination coefficient for the Burr-Hatke-exponential distribution model is larger than the exponential distribution model and the inverse-exponential distribution model.So, the Burr-Hatkeexponential distribution model can be regarded as an efficient model in terms of goodness-of-fit.However, all of the proposed models have an estimated value of more than 90%, so all models are judged as efficient models.Through this study, it is concluded that software operators using the mean square error, the mean value function and the pattern of the hazard function as a preliminary information for detect the software failure phenomenon can be developed the failure type of the lifetime distribution.