Modified moment estimators based on non-conventional measures for the power function distribution

Abstract

Power function is amougst the most suitable probability models for survival or failure times analysis, particularly of electronic components and product reliability. The article proposes some new modified moment estimators for parameter estimation of the power function distribution. The proposed estimators are based on some non-conventional descriptive measures like harmonic mean, quartile deviation, Shannon entropy and Gini index. The performance of the proposed estimators is compared with the traditional moment and existing modified moment estimators. Performance is assessed through the Monte Carlo simulation and three real-life data sets representing failure and survival times of components and infected animals, respectively. Some common accuracy measures are used as performance indicators. From both, Monte Carlo simulation and all real-life applications, the results show better performance of proposed modified moment estimators based on the Gini index and harmonic mean. Hence, the use of these modified estimators is recommended for parameter estimation of the power function distribution.