Statistica

Comparison of Estimation Methods of the Power Generalized Weibull Distribution

Sanku Dey, Mazen Nassar, Sajid Ali, Devendra Kumar, Enayetur Raheem — Thu, 29 Feb 2024 00:00:00 +0100

This article aims to discuss different estimation methods for the power generalized Weibull distribution. An extensive simulation study is carried out to assess the effectiveness of the estimation of model parameters using numerous well known classical methods of estimation. Furthermore, the Bayes estimators of the unknown parameters are also obtained under different loss functions. Monte Carlo simulations are used to assess the performances of the proposed estimators. Besides, bootstrap/ credible intervals are obtained based on considered methods of estimation. Finally, the potentiality of the distribution is illustrated by means of re-analyzing one real data set.

Discrete New Generalized Pareto Distribution

Kuttan Pillai Jayakumar, Jiji Jose — Thu, 29 Feb 2024 00:00:00 +0100

In this paper we propose a discrete analogue of New Generalized Pareto distribution as a new discrete model using general approach of discretization of continuous distribution. The structural properties of the new distribution are discussed. The shape properties, moments, median, infinite divisibility and stress-strength properties are derived. Estimation of parameters are done using maximum likelihood method. An application of real data set shows the suitability of the proposed model.

On Some Characterizations of the Extended Generalised Shifted Lindley Distribution

Sourav Rana, Saran Ishika Maiti, Arindom Chakraborty — Thu, 29 Feb 2024 00:00:00 +0100

In this article, we unravel an extension of shifted version of Lindley distribution, termed as extended generalized shifted Lindley (EGSL) distribution. Stochastic ordering, moment generating function, reliability characteristics and other relevant properties are studied for this distribution. To estimate the parameters involved, method of maximum likelihood is performed. A detailed simulation study for several choices of parameters is executed as well. Finally, as a comparative exploration, possible fitting of the proposed distribution to a real data along with model fitting by other competent distributions is documented through the aid of a few model checking criteria.

On the Shifted Hybrid Log-Normal Distribution

Damodaran Santhamani Shibu, Soman Latha Nitin, Muhammed Rasheed Irshad — Thu, 29 Feb 2024 00:00:00 +0100

The log-normal distribution is widely used to model positive valued data in many areas of applied research. However, sometimes the log-normal distribution does not completely satisfy the fitting expectations in every real life situations. In this paper, we introduce, investigate, and discuss a more flexible shifted hybrid log-normal distribution for which the log-normal distribution is a special case. Also, various properties, special cases and estimation procedure of the new distribution are discussed. Moreover, the performances of maximum likelihood estimators of the parameters are examined using a brief simulation study. The flexibility and performance of the newdistribution is also illustrated through two applications by fitting two real datasets of different situations.

The Exponentiated Gumbel Lomax Distribution: Properties and Applications

Uchenna Ugwunnaya Uwadi, Elebe Emmanuel Nwezza, Chukwuemeka Onwuzuruike Omekara — Thu, 29 Feb 2024 00:00:00 +0100

A new five-parameter distribution called exponentiated Gumbel Lomax (EGuL) is proposed and studied. The proposed distribution has reverse J-shaped, inverted bathtub-shaped and J-shaped hazard rate function making it suitable for modeling survival and lifetime data. The density of the new distribution is expressed as a linear combination of the exponentiated density of the Lomax distribution. We derive the explicit expression for the quantile function, moments, incomplete moment, moment of residual life, entropy and order statistics of EGuL distribution. The estimation of the parameters of the new model is done using the method of maximum likelihood. A simulation study is employed to ascertain the performance of the maximum likelihood estimates. Two applications are used to illustrate that the new distribution provides a better fit compared to other distributions with the same baseline.