Comparison of estimation methods for the Marshall-Olkin extended Lindley distribution
MetadataShow full item record
The aim of this paper is to compare the parameters' estimations of the Marshall-Olkin extended Lindley distribution obtained by six estimation methods: maximum likelihood, ordinary least-squares, weighted least-squares, maximum product of spacings, Cramer-von Mises and Anderson-Darling. The bias, root mean-squared error, average absolute difference between the true and estimate distributions' functions and the maximum absolute difference between the true and estimate distributions' functions are used as comparison criteria. Although the maximum product of spacings method is not widely used, the simulation study concludes that it is highly competitive with the maximum likelihood method.