Abstract
This paper introduces a truncated Weibullexponential distribution and provides a thorough insight into its mathematical characteristics. These characteristics include moments, generating functions, inverse distribution function, and entropy. Various measures are also discussed about the distribution's reliability. A simulation study is carried out to assess the stability and consistency of the maximum likelihood estimates of the parameters. Finally, two social sciences data sets are used to assess the distribution's relevance in modeling realworld situations.
Similar content being viewed by others
Introduction
As the need to model realworld data problems continuously grows, researchers are expanding classical distributions to meet these new demands. To this end, several authors have proposed extensions to the existing families or classes of distributions, including the generalization of the parent distributions and introducing new models with some additional parameters. This ongoing development of the distribution theory reflects a continuous effort to improve the accuracy and applicability of the statistical models in a wide range of fields. Gupta et al.^{1} introduced a new "Exponentiated GClass of distributions", which involves exponentiating the cumulative distribution function (cdf) to some positive power. Since then, many researchers have proposed different classes of distributions, including Marshall Olkin G–family by Marshall and Olkin^{2}, the beta generalizedG family by Eugene et al.^{3}, the Kumaraswamy–G family by Cordeiro and Nadarajah^{4}, and the exponentiated generalized–G family by Cordeiro et al.^{5}, a general method of generating continuous distributions by Alzaatrh et al.^{6}, the Kumaraswamy Marshall Olkin–G family by Alizadeh et al.^{7}, the transmuted exp–generalized–G family by Yousuf et al.^{8}, and Kum–Transmuted–G family by Afify et al.^{9}. These families of distributions provide several flexible and versatile models that can capture a wide range of data patterns and behaviors. Recently, Alzaatreh et al.^{10} have proposed a method to generate the families of distributions by using the truncated distribution and named the families of distributions as the truncated families of distributions. These families of distributions contain four different ways to generate the truncated families of distributions.
The Weibull distribution, a power transformation of the exponential distribution, was first studied by Fisher and Tippett^{11} in the context of the limiting distribution of extreme values in a sample. Since then, the Weibull distribution has been used in various applications, such as modeling the variability in diameter of powder particles by De Moivre^{12} and the failure rate properties of a system over time by Yong^{13}. The Weibull distribution has been used to generate several new model for modeling of more complex phenomena. For instance, Alizadeh et al.^{14} have introduced the Transmuted Weibull–G family of distributions, Tahir et al.^{15} have derived a Weibull–G family of distributions, Cordeiro et al.^{16} have derived the exponentiated Weibull–G distributions, and Oluyede^{17} has proposed the Gamma Weibull–G family of distributions. These developments have expanded the applicability of the Weibull distribution and have provided more flexible models to capture a wider range of data patterns and behaviors.
Statistical models have been expanded to handle the complexity that arises in lifetime analysis. In certain situations, the domain of application of a phenomenon is reduced and one has to restrict the domain of the applicability of the underlying model. The truncated distributions play a useful role in such situations. The truncated distributions can be viewed as a form of a conditional distribution where the domain of the random variable is restricted under some conditions. When the domain of the random variable is restricted below a certain threshold, the resulting distribution is known as the left–truncated distribution. Conversely, if the domain is restricted above a specific threshold, the resulting distribution is a righttruncated distribution.
In this article, we have proposed a new truncated distribution called the truncated Weibullexponential distribution, which is a truncated mixture of Weibull and exponential distribution. The proposed model is flexible and can be used for modeling data from various fields, including reliability sciences, social sciences, demography, and environmental sciences. We believe that the truncated Weibullexponential distribution will provide a valuable addition to the existing models and will serve as an effective tool for data analysis in various fields.
The article structure follows. Section "Methodology" outlines the methodology used in this study. Section "The truncated Weibull–exponential distribution" presents the development of the truncated WeibullExponential model. Section "Properties of the distribution" contains important properties of the proposed distribution. Section "Reliability analysis" discusses some reliability measures. Section "Estimation" contains estimation of the model parameters. Section "Simulation study" is based on the simulation study to see the consistency of the estimation method. Some real data applications are given in Section "Applications". Section "Conclusions" contains the conclusions and findings of the study.
Methodology
Mahdavi and Silva^{18} have derived a truncated family of distributions that can be used with any baseline distribution. The cumulative distribution function (cdf) and probability density function (pdf) of this new family are
and
where \(H_{T} \left( t \right)\) is cdf of any generator distribution, \(H_{T} \left( 0 \right)\) and \(H_{T} \left( 1 \right)\) are values of \(H_{T} \left( t \right)\) at 0 and 1, \(F\left( x \right)\) is cdf of any baseline distribution and \(f\left( x \right)\) is the pdf corresponding to \(F\left( x \right)\). Several distributions can be proposed by using different combinations of \(H_{T} \left( t \right)\) and \(F\left( x \right)\). Najarzadegan et al.^{19} have used the family (1) to propose a two parameter truncated WeibullG family of distributions. Bantan et al.^{20} have proposed a truncated BurrG family of distributions by using a Burr generator in (1). Almarashi et al.^{21} have used the Muth distribution as a generator in (1) to propose a truncated MuthG family of distributions. In this paper, we have proposed a modified truncated WeibullF family of distributions by using the following Weibull cdf and pdf as \(H_{T} \left( t \right)\) and \(h_{T} \left( t \right)\) in (1) and (2), respectively,
and
where \(\gamma\) is the scale parameter and k is the shape parameter. Now, using (3) in (1), the cdf of the proposed truncated Weibull–F family of distributions is
The density function corresponding to (5) is
Various distributions can be derived by using different baseline distributions in (5) and (6). In the following, we have derived the truncated Weibull–Exponential distribution by using the exponential baseline distribution in (5) and (6).
The truncated Weibull–exponential distribution
In this section, we have proposed the truncated Weibull–Exponential (TWEx) distribution by using the following pdf and cdf of the exponential distribution in (5) and (6).
\(F_{X} \left( x \right) = 1  \exp \left( {{{  x} \mathord{\left/ {\vphantom {{  x} \lambda }} \right. \kern0pt} \lambda }} \right)\,\,{\text{and}}\,\,f_{X} \left( x \right) = \left( {{1 \mathord{\left/ {\vphantom {1 \lambda }} \right. \kern0pt} \lambda }} \right)\exp \left( {{{  x} \mathord{\left/ {\vphantom {{  x} \lambda }} \right. \kern0pt} \lambda }} \right)\,\,;\,\,x,\lambda > 0\).
The cdf of the proposed TWEx distribution is
The density function corresponding to (8) is
It is interesting to note that if random variable X has density (8) then the distribution of \(Y = \sqrt X\) has WeibullRayleigh distribution, given by Khalifa et al.^{22}. The TWEx distribution provides some specific distributions as a special case. For example, for k = 1, the distribution reduces to a truncated exponential–exponential (TEEx) distribution, and for k = 2, it reduces to the truncated Rayleigh–exponential (TREx) distribution. It is to be noted that the proposed model is applicable for positive phenomenon only.
The density function of TWEx distribution can be easily expressed as the weighted sum of the exponentiatedexponential distributions as
where \(B_{i,j} \left( {k,\gamma } \right) = \frac{{\left( {  1} \right)^{i + 1} \Gamma \left[ {2k\left( {j + 1} \right)} \right]}}{{i!j!\left( {1  {\text{e}}^{{  \gamma^{  k} }} } \right)\Gamma \left[ {k\left( {j + 1} \right)  1} \right]\gamma^{kj + 1} }}\).
The plots of the density function of TWEx distribution for specific values of λ and γ and for different choices of the parameter k are given in Fig. 1, below
From the above figure, we can see that the shape of the distribution changes with the value of k. We can see that the distribution approaches a symmetrical distribution for a large value of k. We can also see that the parameters \(\lambda\) and \(\gamma\) do not have a significant effect on the shape of the distribution. These parameters rather control the spread of the distribution.
Shape of the distribution
The shape of the distribution is important in certain applications. In the following, we will obtain the mode of the distribution. For this, we first see that the logarithm of the density function is
The derivative of (10) with respect to x is
The mode is obtained as a solution of \({{\partial g_{X} \left( x \right)} \mathord{\left/ {\vphantom {{\partial g_{X} \left( x \right)} {\partial x}}} \right. \kern0pt} {\partial x}} = 0\) and a unique mode of the distribution is readily obtained as
The mode can be computed for specific values of the parameters.
Properties of the distribution
In this section, some desirable properties of the proposed TWEx distribution are explored. These properties are studied in the following subsections.
The quantile function
The quantile function is useful in obtaining percentiles of the distribution. The quantile function is obtained by solving \(G\left( x \right) = m\) for x. Now, for the TWEx distribution, the quantile function is obtained by solving
for x and is given as
The percentiles can be obtained by using any value of m, such that \(m \in \left[ {0,1} \right]\), in (11). Specifically, the median can be obtained by using m = 0.5 in (11). Also, a random observation can be drawn from the TWEx distribution by replacing m with a uniform random number between [0,1] in (11).
Moments and generating functions
The moments are useful in studying certain properties of a probability distribution. The rth moment of a random variable X having cdf G(x) is computed as
\(\mu_{r}^{/} = \int_{  \infty }^{\infty } {x^{r} dG_{X} \left( x \right)} = \int_{  \infty }^{\infty } {x^{r} g_{X} \left( x \right)dx} .\)
Now, using the density function of the TWEx distribution, from (9), in the above equation, we have
Solving the integral, the rth moment of the TWEx distribution is
The moments can be computed for specific values of r in (12).
The mean and variance of the distribution, for selected values of the parameters, are given in Table 1, below
From the table, we can observe that as the value of \(\lambda\) increases, the mean of the TWEx distribution also increases. This is because \(\lambda\) is the scale parameter of the distribution, and as it increases, the distribution shifts towards higher values, resulting in a larger mean. Similarly, we can see that as the parameter \(\gamma\) increases, the mean of the TWEx distribution decreases. This is because \(\gamma\) controls the rate at which the distribution decays, and as it increases, the distribution becomes more concentrated around lower values, resulting in a smaller mean. Finally, as k increases, the mean of the TWEx distribution also increases. This is because k is a shape parameter that controls the rate at which the tails of the distribution decay, and as it increases, the distribution becomes less heavytailed, resulting in a larger mean. The pattern for variance is almost the same as that of the mean.
The incomplete moment is another meaningful measure of a distribution. The incomplete moment of a distribution is defined as
Using (9), the incomplete moment for the TWEx distribution is given as
Simplifying the integral, the pth incomplete moment for the TWEx distribution is
where \(\Gamma \left( {n,z} \right)\) is the incomplete gamma function defined as
The incomplete moments can be computed for specific values of y and p.
The moment generating function (mgf) is a useful function that can be used to compute moments of a distribution. The moment generating function of a continuous random variable X having density \(f\left( x \right)\) is obtained as
Now, the mgf for the TWEx distribution is obtained as
Solving the integral, the mgf of the TWEx distribution is
The moments can be computed from (13).
Entropy
The entropy is a useful measure of information. The Rényi^{23} entropy is computed as
Using the density function of the TWEx distribution, the Rényi entropy is
Transforming we have
Solving the integral, the entropy is
The entropy can be computed for given values of the parameters.
Reliability analysis
In this section, we have given some reliability analysis for the proposed TWEx distribution. The reliability analyses include the survival function, the hazard rate function, the moments of the residuals and the reversed residuals. These are given in the following subsections.
Survival Function
The survival function is useful in reliability analysis. The function is used to obtain the probability that a component will be functioning after a specified time. The survival function is computed as \(S\left( x \right) = 1  F\left( x \right)\). The survival function for the TWEx distribution is given as
The survival function can be computed for specific values of the parameters.
Hazard rate function
The hazard rate function provides the information about instantaneous failure of a component that is still functioning at a specific time point. The hazard rate function for the TWEx distribution is obtained as
The plots of the hazard rate function for different values of the parameters are given in Fig. 2 below
We can see that the hazard rate function of TWEx distribution has both increasing and decreasing trends.
Moments of residual and reversed residual life
The residual life is useful to determine the life expectancy of components. The moments of residual and reversed residual life are useful in obtaining certain information about the life expectancy of the components. The rth moment of residual life is given as
where \(R\left( t \right)\) is survival function associated with \(g_{X} \left( x \right)\) at time t. Using the binomial expansion of \(\left( {x  t} \right)^{r}\), we have
Now, using \(g_{X} \left( x \right)\) from (9), the rth moment of residual life for the TWEx distribution is
Solving the integral, the rth moment of residual life from the TWEx distribution is
The mean residual life can be obtained by using r = 1 in the above equation. The rth moment of reversed residual life is given as
Using the density function, given in (9), the rth moment of the reversed residual life for the TWEx distribution is given as
The mean residual life can be obtained by using r = 1 in (14).
Estimation
In this section, we have discussed the maximum likelihood estimation of parameters of the TWEx distribution. For this, suppose that a sample of size n is available from the TWEx distribution. The loglikelihood function is then
The derivatives of the loglikelihood function, (15), with respect to the unknown parameters are
and
The maximum likelihood estimators of k, γ and \(\lambda\) can be obtained by equating the derivatives in (16–18) to zero and numerically solving the resulting equations.
The Fisher information matrix is useful to obtain the variances and covariances of the maximum likelihood estimators of the parameters. The entries of the Fisher information matrix for the TWEx distribution are given as
The entries of the Fisher information matrix are given in Appendix–A. The inverse of the Fisher information matrix provides the variance–covariance matrix of the maximum likelihood estimates, which help in obtaining the confidence intervals for the true population parameters.
Simulation study
In this section, we have presented the simulation study to see the performance of the maximum likelihood estimates of the parameters. The simulation study has been conducted by generating random samples of different sizes from the TWEx distribution. The simulation algorithm is given below:

1.
Generate random samples of sizes 50, 250, 500, 750, and 1000 from the TWEx distribution.

2.
Compute maximum likelihood estimates of the parameters \(\gamma ,\lambda\) and k for each sample.

3.
Repeat steps 1 and 2 for 10,000 times.

4.
Compute average estimates and the mean square error as
\(\overline{\hat{\theta }} = \frac{1}{10000}\sum\limits_{i = 1}^{10000} {\hat{\theta }_{i} } \,\,;\,\,MSE\left( {\hat{\theta }} \right) = \frac{1}{10000}\sum\limits_{i = 1}^{10000} {\left( {\hat{\theta }_{i}  \overline{\hat{\theta }}} \right)^{2} } \,\,;\,\,\hat{\theta } = \hat{\gamma }\,,\,\,\hat{\lambda }\,\,,\,\,\hat{k}\) and \(\overline{\hat{\theta }} = \frac{1}{10000}\sum\limits_{i = 1}^{10000} {\hat{\theta }_{i} }\).
The results of the simulation study are given in Table 2 below; with prespecified values in the parenthesis;
From the above table, we can see that the estimated values of the parameters are close to the prespecified values. Also, we can see that the mean square error of the estimates decreases with an increase in the sample size. The results of simulation study indicate that the maximum likelihood estimates of the parameters are consistent estimates.
Applications
In this section, we have given two real data applications of the proposed TWEx distribution. The data set has been obtained from the Country Data Book^{24} and contains information about poverty level and college degrees. The summary measures of two data are given in Table 3, below
We have fitted the following distributions; alongside the TWEx; distribution on these two data sets

Weibull (W): \(g_{X} \left( x \right) = \frac{k}{\lambda }\left( {\frac{x}{\lambda }} \right)^{k  1} \exp \left[ {  \left( {\frac{x}{\lambda }} \right)^{k} } \right]\,\,;\,\,x,k,\lambda > 0\)

Exponential (E): \(g_{X} \left( x \right) = \frac{1}{\lambda }\exp \left( {  \frac{x}{\lambda }} \right)\,\,;\,\,x,\lambda > 0\)

Log–Logistic (LL): \(g_{X} \left( x \right) = \frac{{\gamma \lambda^{\gamma } x^{\gamma  1} }}{{\left( {\lambda^{\gamma } + x^{\gamma } } \right)^{2} }}\,\,;\,\,x > 0\,\,,\,\,\lambda ,\gamma > 0\)

Exponentiated Weibull (EW): \(g_{X} \left( x \right) = \frac{k\gamma }{\lambda }\left( {\frac{x}{\lambda }} \right)^{\gamma  1} \exp \left[ {  \left( {\frac{x}{\lambda }} \right)^{\gamma } } \right]\left[ {1  \exp \left\{ {  \left( {\frac{x}{\lambda }} \right)^{\gamma } } \right\}} \right]^{k  1}\)
The estimated model parameters alongside the values of − 2(loglikelihood), Akaike Information Criterion (AIC), and the p value of Kolmogorov–Smirnov (KS) test are given in Table 4 below
From the above table, we can see that the TWEx distribution is the best fit for the two data sets as it has the smallest value of AIC and the largest p value for the KS–test among all the competing distributions.
We have also constructed the plots of density and distribution functions of various distributions for the two data sets. These plots are given in Figs. 3 and 4 below. From these plots, we can see that the TWEx distribution is the best fit for the two data sets. Also, the loglogistic distribution is second best fit to the data. The other distributions used in the analyses are the poor fit. Further, the probability plot of the TWEx distribution for the two data sets, given in Fig. 5, shows the same result that the TWEx distribution is a reasonable fit for the two data sets.
Conclusions
In this paper, we have proposed a TWEx distribution using the truncated Weibull distribution as a generator. Various desirable properties of the proposed TWEx distribution are explored. These properties include moments of the distribution, quantile function, entropy, and generating function. We have also discussed the reliability analysis for the proposed distribution. It has been observed that the proposed distribution has increasing and decreasing hazard rate functions. The moments of residual and reversed residual life has also been computed. We have also discussed the maximum likelihood estimation of the parameters for the proposed TWEx distribution. A simulation study is presented to see the performance of the estimates and it is found that the maximum likelihood estimates are consistent as the estimates converges to the true parameter values. Two reallife data applications of the proposed distribution have also been discussed and it is found that the proposed distribution is the best fit for the data in comparison with the other competing distributions.
Data availability
The data is available with the authors and can be obtained on request. Also, no medicine was tested on humans and animals for data analysis in this study.
References
Gupta, R. C., Gupta, P. I. & Gupta, R. D. Modeling failure time data by Lehmann alternatives. Commun. Stat.: Theory Methods 27, 887–904 (1998).
Marshall, A. & Olkin, I. A new method for adding a parameter to a family of distributions with applications to the exponential and Weibull families. Biometrika 84, 641–652 (1997).
Eugene, N., Lee, C. & Famoye, F. Betanormal distribution and its applications. Commun. Stat.: Theory Methods 31, 497–512 (2002).
Cordeiro, G. M. & Nadarajah, S. Closedform expressions for moments of a class of beta generalized distributions. Braz. J. Probab. Stat. 25, 14–33 (2011).
Cordeiro, G. M., Ortega, E. M. M. & da Cunha, D. C. C. The exponentiated generalized class of distributions. J. Data Sci. 11, 1–27 (2013).
Alzaatreh, A., Lee, C. & Famoye, F. A new method for generating families of continuous distributions. Metron 71, 63–79 (2013).
Alizadeh, M., Cordeiro, G. M., Brito, C. & Demetrio, C. G. The beta MarshallOlkin family of distributions. J. Stat. Dist. Appl. 4, 1–18 (2015).
Yousof, H. M. et al. The Transmuted Exponentiated Generalized–G Family of Distributions. Pak. J. Stat. Op. Res. 11, 441–464 (2015).
Afify, A. Z., Cordeiro, G. M., Yousof, H. M., Alzaatreh, A. & Nofal, Z. M. The Kumaraswamy transmutedG family of distributions: Properties and applications. J. Data Sci. 14, 245–270 (2016).
Alzaatreh, A. et al. Truncated family of distributions with applications to time and cost to start a business. Methodol. Comput. Appl. Probab. 23, 5–27 (2021).
Fisher, R. A. & Tippett, L. H. C. Limiting forms of the frequency distribution of the largest or smallest member of a sample. Math. Proc. Cambridge Philos. Soc. 24, 180–190 (1928).
De Moivre, A. The Doctrine of Chances: A Method of Calculating the probability of events in Play. (Pearson, 1718).
Yong, T. Extended Weibull distributions in reliability engineering, PhD Thesis submitted to the National University of Singapore, (2004).
Alizadeh, M., Rasekhi, M., Yousof, H. M. & Hamedani, G. G. The transmuted WeibullG family of distributions. Hacet. J. Math. Stat. 47, 1671–1689 (2017).
Tahir, M. H. et al. A new WeibullG family of distributions. Hacet. J. Math. Stat. 45, 629–647 (2016).
Cordeiro, G. M., Afify, A. Z., Yousof, H. M., Pescim, R. R. & Aryal, G. R. The exponentiated WeibullH family of distributions: Theory and applications. Mediterr. J. Math. 14, 155. https://doi.org/10.1007/s0000901709551 (2017).
Oluyede, B. The gammaWeibullG family of distributions with applications. Austrian J. Stat. 47, 45–76 (2018).
Mahdavi, A. & Silva, O. G. A method to expand family of continuous distributions based on truncated distributions. J. Stat. Res. Iran 13, 231–247 (2016).
Najarzadegan, H., Alamatsaz, M. H. & Hayati, S. Truncated WeibullG more flexible than betaG distribution. Int. J. Stat. Probab. 6, 1–17 (2017).
Bantan, R. A. R., Chesneau, C., Jamal, F., Elbatal, I. & Elgarhy, M. The truncated BurrG family of distributions: Properties and applications to actuarial and financial data. Entropy 23, 1088 (2021).
Almarashi, A. M., Jamal, F., Chesneau, C. & Elgarhy, M. A new truncated muth generated family of distributions with applications. Complexity 2021, 1211526 (2021).
Khalifa, E. H., Ramadan, D. A. & ElDesouky, B. S. Statistical inference of truncated WeibullRayleigh distribution: Its properties and applications. Open J. Model. Simul. 9, 281–298 (2021).
Rényi, A. On measures of information and entropy. In Proceedings of the Fourth Berkeley Symposium on Mathematical Statistics and Probability, 547–561 (1961).
County and City Data Book, 12th Edn, U.S Department of Commerce.
Author information
Authors and Affiliations
Contributions
S.A., M.F., and S.H.S. obtained the mathematical results and wrote the initial draft. S.A. and J.A.D. conducted a simulation study and real data analysis. M.Q.S. provided the main idea, validated all the results, and finalized the draft.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix A
Appendix A
Entries of Fisher Information Matrix for TWEx distribution. We have assumed \(1  {\text{e}}^{{  \gamma^{  k} }} = z\) and \(1  {\text{e}}^{{{{  x_{i} } \mathord{\left/ {\vphantom {{  x_{i} } \lambda }} \right. \kern0pt} \lambda }}} = w_{i}\)
and
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Abbas, S., Farooq, M., Darwish, J.A. et al. Truncated Weibull–exponential distribution: methods and applications. Sci Rep 13, 20849 (2023). https://doi.org/10.1038/s4159802348288x
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s4159802348288x
This article is cited by

Global population: from SuperMalthus behavior to Doomsday criticality
Scientific Reports (2024)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.