Power new generalized class of Kavya–Manoharan distributions with
an application to exponential distribution · 2025
Read Power new generalized class of Kavya–Manoharan distributions with by an application to exponential distribution (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-18930.txt.
Introduction
Lazhar Benkhelifa extends a young family of probability curves with a power twist, in a 2025 paper. The Kavya-Manoharan class, introduced the same year, already serves reliability work and lifetime modeling. Adding a power step stretches the class further, letting one formula bend into many shapes.
Taking the exponential curve as the base, the new model handles both steadily rising-or-falling and hump-shaped hazard rates, and it contains eleven named submodels. The paper derives its working traits: moments, quantiles, entropy, and reliability measures. Parameters are fitted by maximum likelihood, simulations check the estimator's honesty, and a real dataset shows the new curve out-fitting its rivals. Flexible lifetimes, one family.
Catalog record
| CREATOR | an application to exponential distribution |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 37 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-18930
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Power new generalized class of Kavya–Manoharan distributions with an application to exponential distribution Abstract Recently, Verma et al. (2025) introduced a novel generalized class of Kavya-Manoharan distributions, which have demonstrated significant utility in reliability analysis and the modeling of lifetime data. This paper proposes an extension of this class by applying the power generalization technique, thereby enhancing more flexibility and applicability. We take the exponential distribution as the baseline distribution to introduce a new model capable of accommodating both monotonic and non-monotonic hazard rate functions. Our model includes eleven submodels. We present several statistical properties of the introduced model, including moments, generating and characteristic functions, mean deviations, quantile function, mean residual life function, Rényi entropy, order statistics, and reliability. To estimate the unknown model parameters, we use the maximum likelihood approach. A simulation study is conducted to assess the validity of the maximum likelihood estimator. The superiority of the new distribution is demonstrated through the use of a real data application. Keywords: Reliability, Lifetime model, Hazard rate, Maximum likelihood estimation. 1 Introduction In reliability, lifetime models are important tools for modeling and analyzing the time until a failure happens. The exponential distribution has long been a popular choice for modeling lifetime data due to its simplicity and ease of use. But it’s a notable limitation when we want to model the data that does not exhibit a constant hazard rate. In order to get rid of this limitation, researchers have suggested several extensions and generalizations of the exponential distribution via transformation methods. The aim of these generalizations is to provide greater flexibility to model various types of data by introducing at least one parameter or transforming the parent distribution. For example, Gupta et al. (1998) used the power generalization technique to introduce the exponentiated exponential distribution. Nadarajah and Kotz (2006) suggested another generalization of the exponential distribution, derived from the logit transformation of a beta random variable, which they termed the beta exponential distribution. Kumar et al. (2015) presented the Dinesh-Umesh-Sanjay (DUS) transformation and studied the DUS exponential distribution. The alpha power transformation (APT), with an application to exponential distribution, was developed by Mahdavi and Kundu (2017). Kavya and Manoharan (2021) introduced the Kavya-Manoharan (KM) transformation method, whereas the power generalized DUS (PGDUS) transformation was proposed by Thomas and Chacko (2021). Lone and Jan (2023) introduced the method known as the pi-exponentiated transformation (PET). The power generalized KM (PGKM) transformation for non-monotone hazard rate was presented by Deepthi and Chacko (2023). More recently, by extending the KM transformation, Verma et al. (2025) developed a new generalized KM (NGKM) transformation technique. They introduced the NGKM exponential distribution utilizing the exponential distribution as the baseline model. The cumulative distribution function (CDF) of the NGKM transformation is given by: F\left(x\right)=\left\{\begin{tabular}[]{ll}$\frac{\lambda}{\lambda-1}\left[1-\lambda^{-G\left(x\right)}\right]$&if $\lambda$ $>0,\lambda\neq 1,$\\ $G\left(x\right)$&if $\lambda$ $=1,$\end{tabular}\ \ \ \ \right. where G(x) is the CDF of the baseline distribution and \lambda is the shape parameter. In this paper, we apply the power generalization technique to propose a new extension of the NGKM transformation, referred to as the power NGKM (PNGKM) transformation. Adding a new shape parameter \alpha>0, we get the CDF of the PNGKM family, which is: F\left(x\right)=\left\{\begin{tabular}[]{ll}$\frac{\lambda^{\alpha}}{\left(\lambda-1\right)^{\alpha}}\left[1-\lambda^{-G\left(x\right)}\right]^{\alpha}$&if $\lambda$ $>0,\lambda\neq 1,$\\ $G^{\alpha}\left(x\right)$&if $\lambda$ $=1.$\end{tabular}\right. (1) Note that the PNGKM transformation contains some transformations techniques as pecial cases: • If \lambda=1 , then we obtain the exponentiated (power) transformation; which is proposed by Gupta et al (1998). • If \alpha=1 and \lambda=e^{-1} , then we get the DUS transformation; see Kumar et al. (2015). • If \alpha=1 and \lambda=\theta^{-1} , then we obtain the APT; see Mahdavi and Kundu 2017. • If \alpha=1 and \lambda=e , then we get the GKM transformation; see Kavya and Manoharan (2021). • If \lambda=e , then we obtain the PGKM transformation; see Deepthi and Chacko (2023). • If \lambda=e^{-1} , then we get the PDUS; see Thomas and Chacko (2023). • When \alpha=1, \lambda=\pi^{-1}\ and \gamma=1 , we obtain the PET; see Lone and Jan (2023). • When \alpha=1 , we obtain the NGKM transformation; see Verma et al. (2025). • When \lambda=\theta^{-1} , we get the power APT (PAPT); new. • When \lambda=\pi^{-1}\ and \gamma=1 we get the power PET (PPET); new. The paper is structured as follows. Section 2 presents the PNGKM exponential distribution. Various statistical properties of our suggested distribution are given in Section 3. The maximum likelihood estimators of the model parameters and the examination of the performance of these estimators through a simulation are presented in sections 4 and 5, respectively. An application to well-known real data is given in Section 6. Finally, Section 7 was devoted to the conclusion.. 2 PNGKM exponential distribution It is known that the CDF of exponential distribution is G\left(x\right)=1-e^{-\beta x}, \beta>0,\ x>0. Then, substituting G\ in \left(1\right) we get the CDF of the PNGKM exponential (PNGKME) distribution as follows: F\left(x\right)=\left\{\begin{tabular}[]{ll}$\frac{\lambda^{\alpha}}{\left(\lambda-1\right)^{\alpha}}\left(1-\lambda^{-\left(1-e^{-\beta x}\right)}\right)^{\alpha}$&if $\lambda$ $>0,\lambda\neq 1,$\\ $\left(1-e^{-\beta x}\right)^{\alpha}$&if $\lambda$ $=1.$\end{tabular}\ \ \ \ \ \ \ \right. (2) The corresponding probability density function (PDF) and the hazard rate function are, respectively, given by f\left(x\right)=\left\{\begin{tabular}[]{ll}$\frac{\alpha\beta\lambda^{\alpha}\ln\lambda}{\left(\lambda-1\right)^{\alpha}}e^{-\beta x}\lambda^{-\left(1-e^{-\beta x}\right)}\left(1-\lambda^{-\left(1-e^{-\beta x}\right)}\right)^{\alpha-1}$&if $\lambda$ $>0,\lambda\neq 1,$\\ $\alpha\beta e^{-\beta x}\left(1-e^{-\beta x}\right)^{\alpha-1}$&if $\lambda$ $=1,$\end{tabular}\ \ \ \ \ \ \ \right. (3) and h\left(x\right)=\left\{\begin{tabular}[]{ll}$\frac{\alpha\beta\lambda^{\alpha}\ln\left(\lambda\right)e^{-\beta x}\lambda^{-\left(1-e^{-\beta x}\right)}\left(1-\lambda^{-\left(1-e^{-\beta x}\right)}\right)^{\alpha-1}}{\left(\lambda-1\right)^{\alpha}-\lambda^{\alpha}\left(1-\lambda^{-\left(1-e^{-\beta x}\right)}\right)^{\alpha}}$&if $\lambda$ $>0,\lambda\neq 1,$\\ $\frac{\alpha\beta e^{-\beta x}\left(1-e^{-\beta x}\right)^{\alpha-1}}{1-\left(1-e^{-\beta x}\right)^{\alpha}}$&if $\lambda$ $=1.$\end{tabular}\ \ \ \ \ \ \ \right. The PNGKME distribution includes as special cases many important well-known distributions: • When \lambda=1 , it reduces to exponentiated exponential (EE) distribution; see Gupta et al (1998). • When \alpha=1 and \lambda=e^{-1} , we get the DUS exponential (DUSE) distribution; see Kumar et al. (2015). • When \alpha=1 and \lambda=\theta^{-1} , we obtain the APT exponential (APTE) ; see Mahdavi and Kundu (2017). • When \alpha=1 and \lambda=e , it reduces to GKME distribution; see Kavya and Manoharan (2021). • When \lambda=e , it reduces to PGKME distribution; see Deepthi and Chacko (2023). • If \lambda=e^{-1} , then we obtain the PDUSE distribution; see Thomas and Chacko (2021). • When \alpha=1, \lambda=\pi^{-1}\ and \gamma=1 , we get the PETE distribution; see Lone and Jan (2023). • When \alpha=1 , we get the NGKME distribution; see Verma et al. (2025). • When \lambda=\theta^{-1} , we obtain the PAPT exponential (PAPTE) distribution; new. • If \lambda=\pi^{-1} and \gamma=1 , then it reduces to PPET exponential (PPETE) distribution; new. • If \lambda=1 and \alpha=1, then we get the exponential distribution. The general shapes of the PDF and hazard rate function for the PNGKME distribution are difficult, or even impossible, to establish analytically. Nonetheless, we can graphically illustrate these shapes for some selected values of \alpha, \beta and \lambda . Figure 1 gives the various shapes of the PDF of the PNGKME distribution. From this figure, it is evident that the PDF has a decreasing or a unimodal shape depending on the parameter values. Similarly, various possible shapes of the hazard function of the PNGKME distribution are given in Figure 2. We observe that the shapes of hazard rates are decreasing, increasing, upside-down bathtub (unimodal), or constant according to the choice of parameters. Figure 1: PDF plot of the PNGKME distribution. Figure 2: Some possible shapes of the hazard function 3 Statistical properties Note that a survey of the properties of the EE distribution for \lambda