Abstract

Common in medical studies or reliability analysis, the failure of individuals or units may be attributable to more than one cause.  Also, design the life-testing experiments under a highly reliable product or materials required a long period of time which requires stress higher than normal stress. In this paper,  we adopt the mixture of two exponential distributions which has become increasingly necessary in engineering statistics. Under a step-stress partially accelerated life tests and type-II censoring scheme for independent competing risks data the model is formulated. The estimation problem is  addressed here from a classical viewpoint. The maximum likelihood estimates of the unknown model parameters are formulated. Additionally, the asymptotic confidence intervals depend on normality theory, along with the two bootstrap confidence intervals are proposed. Finally, to validate the proposed model and assess the estimation methods, we formulated an  extensive simulation study.

Statistics