Inference and quantile regression for the unit-exponentiated Lomax distribution

Aisha Fayomi; Amal S. Hassan; Ehab M. Almetwally

doi:10.1371/journal.pone.0288635

Abstract

In probability theory and statistics, it is customary to employ unit distributions to explain practical variables having values between zero and one. This study suggests a brand-new distribution for modelling data on the unit interval called the unit-exponentiated Lomax (UEL) distribution. The statistical aspects of the UEL distribution are shown. The parameters corresponding to the proposed distribution are estimated using widely recognized estimation techniques, such as Bayesian, maximum product of spacing, and maximum likelihood. The effectiveness of the various estimators is assessed through a simulated scenario. Using mock jurors and food spending data sets, the UEL regression model is demonstrated as an alternative to unit-Weibull regression, beta regression, and the original linear regression models. Using Covid-19 data, the novel model outperforms certain other unit distributions according to different comparison criteria.

Citation: Fayomi A, Hassan AS, Almetwally EM (2023) Inference and quantile regression for the unit-exponentiated Lomax distribution. PLoS ONE 18(7): e0288635. https://doi.org/10.1371/journal.pone.0288635

Editor: Qichun Zhang, University of Bradford, UNITED KINGDOM

Received: December 28, 2022; Accepted: July 1, 2023; Published: July 18, 2023

Copyright: © 2023 Fayomi et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The data inside the paper.

Funding: The authors received no specific funding for this work.

Competing interests: The authors have declared that no competing interests exist.

1 Introduction

Lomax [1] established the Lomax (L) or Pareto II distribution to model data on business failure. This distribution has found widespread use in a variety of domains, including income and wealth disparity [2, 3], city size [4], the size distribution of computer files on servers [5], internet traffic [6], and receiver operating characteristic curve analysis [7]. Reference [8] claimed that the Lomax distribution is a good fit for modeling reliability issues because many of its characteristics can be understood in that context and could serve as an alternative to the well-known distributions used in reliability. Reference [9] utilized this distribution to model size spectrum data in aquatic ecology. Many descriptions are provided for the characterization of the Lomax distribution. It is referred to as a particular model of the Pearson Type VI distribution. It has also been thought of as a combination of the exponential and gamma distributions. The L model emerges as a limiting distribution of residual lifetimes at very old ages [10]. The L belongs to the family of decreasing failure rates in the context of lifetimes [11]. Reference [12] mentioned that the L distribution is a heavy-tailed alternative to the exponential, Weibull, and gamma distributions. Moreover, it belongs to the Burr family of distributions [13].

The cumulative distribution function (CDF) of the Lomax distribution is defined by: (1) where, λ, δ ∈ R⁺, are the scale and shape parameters respectively. The probability density function (PDF) of the Lomax distribution is as follows: (2) From PDF (2), it is clear that the L distribution simplifies to:

The beta prime (or inverted beta) distribution for λ = 1, δ ≠ 1.
The log-logistic distribution for δ = 1, λ ≠ 1.
F(2,2) distribution for δ = λ = 1.

To gain a better match for data analysis in different disciplines, modified and expanded variants of the L distribution were established. The next are a few of these broad generalizations that are highlighted; expoentiated L [14], Marshall Olkin L distribution [15], McDonald L distribution [16], Weibull L distribution [17], gamma L distribution [18], exponentiated Weibull L distribution [19], Maxwell L distribution [20], and Nadarajah-Haghighi L distribution [21].

The interest in the present study with the exponentiated L (EL) distribution with an extra shape parameter ϑ compared to (1). The EL distribution has the following CDF and PDF: (3) (4) where Θ ≡ (δ, λ, ϑ) ∈ R⁺. The EL model provides well-known distributions for certain specific values of parameters. It includes Lomax distribution for ϑ = 1, and exponentiated Pareto distribution for λ = 1.

Recently, the development of new flexible probability distributions to provide well-fitting models to datasets with values ranging from 0 to 1, has piqued statisticians’ curiosity. These bounded distributions are necessary for modeling proportions, percentages, and probabilities. In applied disciplines, there is a strong need for the analysis of datasets on the (0, 1) for semi-parametric or parametric and regression models. the analysis of datasets on the (0, 1) for parametric, semi-parametric, and regression models is also in high demand. Furthermore, unit distributions allow extra flexibility throughout the unit interval without introducing new parameters to the fundamental distribution. The following are some of the most important unit distributions with diverse numbers of parameters: log-Lindley distribution [22], unit-Birnbaum-Saunders distribution [23], unit-Gompertz distribution [24], unit-inverse Gaussian distribution [25], unit-Weibull distribution [26], unit-Burr-XII distribution [27], unit-Gamma/Gompertz distribution [28], unit exponentiated half logistic distribution [29], unit power Burr X distribution [30] and unit inverse exponentiated Weibull distribution [31].

In this study, in light of the above, an inverse-exponential transformation is used to generate a new unit-flexible probability distribution with a three-parameter based on the EL distribution. The new distribution, named the unit exponentiated Lomax (UEL) distribution, may be utilized to evaluate a wide range of datasets having a value from zero to one. A quantile regression model is developed based on the parametrization of the UEL distribution in terms of the μ^th quantile. In light of the following facts, the UEL distribution is introduced:

To provide a novel distribution defined on (0,1) as a competitor to the existing bounded distributions;
The new distribution can take different hazard rate shapes, such as constant, increasing, decreasing, and bathtub;
It may be considered a model that is appropriate for fitting skewed data that may not be well fitted by other common distributions and can be used to address a wide range of issues in many fields;
To investigate key statistical features of the UEL distribution, such as entropy measures, probability-weighted moments (PWMs), moments, stress-strength (S-S) reliability, and incomplete moments (IM);
To investigate inferential features of the UEL distribution parameters using widespread estimation methodologies, such as maximum likelihood (ML), a maximum product of spacing (MPS), and Bayesian;
To examine the performance of the parameters using a simulation methodology;
Three real-data applications are examined: the first and second data sets are concerned with quantile regression modeling, while the third data set is concerned with data modeling.

The following is a description of the paper’s structure. Section 2 defines the suggested distribution. Section 3 describes its essential distributional features. The methodologies for estimating unknown parameters using various estimation approaches are covered in Section 4. In Section 5, a simulation analysis is conducted to assess the parameter estimates. The new quantile regression model based on the newly specified distribution is described in Section 6. Two real-world applications employing the suggested quantile regression model alongside additional famous regression models are shown in Section 7. The application of the UEL distribution to Covid-19 data in Section 7 reveals that it is preferred to the other seven unit distributions. In Section 8, the paper comes to a close.

2 Model description

In this section, a restriction of the EL distribution in the unit interval is done to introduce a new bounded distribution with support on (0, 1) referring to the UEL distribution. Several extension\ modifications have been done using some transformation, in the present work, a similar exponential transformation is used as provided in References [24–31].

Suppose that Y = e^−X, where X is the EL distribution, then the CDF of the UEL distribution is determined as follows:

Based on the previous equation, the CDF and PDF of the UEL distribution are provided in the following definition.

Definition: A random variable Y is said to follow the UEL distribution with a set of parameters Θ ≡ (δ, λ, ϑ) ∈ R⁺, if it’s CDF and PDF, are given by: (5) and (6)

Note that, F(y; Θ) = 0, for y ≤ 0, and F(y; Θ) = 1, for y > 1. The survival function and hazard rate function (HRF) are as follows:

The density and HRF plots of the UEL distribution are given in Fig 1 for specific values of parameters.

Download:

Fig 1. Density and HRF plots with different parameter values.

https://doi.org/10.1371/journal.pone.0288635.g001

From Fig 1, it can be concluded that the UEL density takes a variety of forms, such as left-skewed, U-shaped, unimodal, and J-shaped. Also, the HRF can be increasing, decreasing, J-shaped, or bathtub-shaped.

3 Characteristics of the UEL distribution

Specific of the UEL distribution’s structural features, such as PWM, moments and IM, quantile function, some entropy measures, and S-S reliability, are investigated in this section.

3.1 Probability weighted moments

The PWM is typically thought to be preferable to ordinary moments. The PWM is less sensitive to extreme values. When ML estimators are difficult to obtain, they are occasionally utilized. The class of PWM, denoted by Ψ_h,s, for a random variable Y, is characterized as follows: (7)

Substituting (5) and (6) in (7), then the PWM of the UEL distribution is

Using the binomial expansion, where s is a positive integer, then Ψ_h,s, is as follows: (8) Using the following binomial expansion in the last term of (8) (9)

Inserting (9) in (8) leads to Let z = −λ ln(y), then the UEL distribution’s PWM is:

Using exponential expansion, in the previous equation, then the PWM of the UEL distribution becomes where B(.,.) stands for beta function (BF).

3.2 Moments & associated measures

If Y has the PDF (6), then the s^th moment is derived as (10)

Let z = −λ ln(y), and use the exponential expansion, then the previous equation is

The s^th central moment, say μ_s, of a given random variable Y, is defined by:

Numerical values for certain parameter values, (a)(λ = 0.5, ϑ = 0.5, δ = 1.5), (b)(λ = 0.5, ϑ = 1.5, δ = 1.5), (c)(λ = 0.5, ϑ = 1.5, δ = 0.5), (d)(λ = 2, ϑ = 0.5, δ = 1.5), (e)(λ = 2, ϑ = 1.5, δ = 1.5), (f)(λ = 2, ϑ = 1.5, δ = 0.5), of the first four moments, variance (σ²), coefficient of skewness (CS) and coefficient of kurtosis (CK) of the UEL distribution are listed in Table 1.

Download:

Table 1. Some moments of the UEL distribution.

https://doi.org/10.1371/journal.pone.0288635.t001

From Table 1, it can be concluded that the moment values decrease with increasing value of ϑ, for fixed values of λ, δ. Also, according to the values of CS and CK, the distribution is right-skewed, platykurtic, and leptokurtic.

3.3 Quantile function

For q ∈ (0, 1), the quantile function (QF) of Y is obtained by inverting (5) as follows: which provides; (11) The first quantile, median and third quantile are obtained, respectively, by setting q = 0.25, 0.5, and 0.75 in (11). It’s simple to simulate the random variable. The random variable Y = y_q at q follows (5), if Q is a uniform variate (0, 1).

3.4 Incomplete moments

The UEL distribution’s s^th lower IM, is given by: Using binomial expansion and let z = −λ ln(y), then η_s(t), can be written as: Also, we use the exponential expansion, then η_s(t), is as follows: Let x = (1 + z)⁻¹, then η_s(t), is as follows: where B(.,.,t) stands for incomplete BF. The Lorenz curve, defined by L(t) is notable applications of the first IM. These curves are particularly important in the fields of economics, demography, insurance, etc.

3.5 Information measures

Here, certain uncertainty measures are investigated including Rényi (Ré) entropy, Havrda and Charvat (H-C) entropy, and ω− entropy.

The Ré entropy of a random variable represents a measure of the variation of the uncertainty. The Ré entropy of a random variable Y has the UEL distribution defined by: (12)

Substituting (6) in (12), then

Using the binomial expansion, then E_R(ω) is converted to

Let z = −λ ln(y), and expand exponentially, then the Ré entropy of the UEL distribution The H-C entropy measure of the UEL distribution is given by The ω− entropy measure of the UEL distribution is obtained as follows: Table 2 gives some numerical values of E_R(ω), H_R(ω), and ξ_R(ω) for the same selected parameter values provided in Table 1.

Download:

Table 2. Entropy measures of the UEL distribution.

https://doi.org/10.1371/journal.pone.0288635.t002

3.6 Stress–strength model

In statistical literature, the term “S-S reliability” is typically represented as A = P[Y₂ < Y₁]. The expression comes from a basic scenario in which a system with random strength Y₁ is exposed to random stress Y₂, with the system failing if the stress exceeds the strength. Let Y₁ and Y₂, are independent random variables with UEL (λ, δ, ϑ₁), and UEL (λ, δ, ϑ₂) distributions. The S-S reliability of the UEL distribution is then determined as follows: (13) Then S-S reliability is obtained as follows

4 Parameter estimation of the UEL model

The parameter estimators of the UEL distribution based on ML, MPS, and Bayesian estimation methods are discussed in this section.

4.1 ML method

Let y₁,…,y_n be the observed samples from the UEL distribution with parameters δ, λ, and ϑ. The likelihood function (LF), say L (y|Θ) of the UEL distribution is expressed as: (14) Then the ln of LF, say ℓ, of the UEL distribution is: (15)

The ML equations, which are based on (15), are therefore as follows: Solving the non-linear equations ∂ℓ/∂δ = 0, ∂ℓ/∂λ = 0, and ∂ℓ/∂ϑ = 0, numerically using optimization algorithm as conjugate-gradient optimization, the ML estimators of δ, λ, and ϑ are obtained.

4.2 MPS method

The MPS technique, which is an alternative to the ML methodology, offers a parameter estimate of a continuous distribution. Suppose that y₍₁₎,…,y_(n) be the observed ordered samples from the UEL distribution with parameters δ, λ, and ϑ. The MPS estimators of δ, λ, and ϑ are generated by maximizing the following: Solving the non-linear equations ∂ℓ(g)/∂δ = 0, ∂ℓ(g)/∂λ = 0, and ∂ℓ(g)/∂ϑ = 0 via numerical technique, the MPS estimators of δ, λ, and ϑ are provided. The MPS estimators are derived by partly differentiating the natural logarithm of the UEL distribution’s product spacing function with respect to population parameters and using an optimization algorithm (conjugate-gradient or Newton-Raphson optimization).

4.3 Bayesian method

Here, the Bayesian estimator of the UEL parameters is produced. The Bayesian estimator is regarded under symmetric (squared error loss function (SELF)) which is defined as Assuming that the prior distribution of δ, λ, and ϑ, denoted by π(δ), π(ϑ), π(λ), have an independent gamma distribution. The joint gamma prior density of δ, λ and ϑ can be written as: (16) From (14) and (16), the joint posterior of the UEL distribution with parameters δ, λ, and ϑ is Bayesian estimators may be produced using the Markov Chain Monte Carlo (MCMC) technique. Gibbs sampling, as well as the more generic Metropolis within Gibbs samplers, are important MCMC techniques. The Metropolis-Hastings (MH) algorithm and Gibbs sampling are two well-known applications of the MCMC approach. To produce random samples from conditional posterior densities of δ, λ and ϑ as follows: The Bayesian estimators are obtained via SELF.

5 Simulation study

A Monte-Carlo simulation study was conducted to analyze and compare the behaviour of different estimates methods based on mean squared errors (MSEs) and biases.

From the UEL distribution, generate 5000 random samples of sizes n = 30, 70, and 150. Various actual parameter values were taken into account as follows: In Table 3: (δ, λ, ϑ) = (0.5, 0.5, 0.5), (0.5, 0.5, 2.5), (0.5, 2.5, 0.5), and (0.5, 2.5, 2.5). In Table 4: (δ, λ, ϑ) = (2.5, 0.5, 0.5), (2.5, 0.5, 2.5), (2.5, 2.5, 0.5), and (2.5, 2.5, 2.5). Since none of the proposed estimators for any unknown parameter can be solved analytically, we can only numerically evaluate them using computing resources and numerical techniques. Because of this, the ML and MPS are calculated using the ‘maxLik’ package, which employs the Newton-Raphson method of maximization in the computations. Additionally, the ‘CODA’ package, which analyses MCMC outputs and diagnoses lack of convergence, is used to compute Bayesian estimation.

Download:

Table 3. The UEL distribution biases and MSEs under various estimation methods at δ = 0.5.

https://doi.org/10.1371/journal.pone.0288635.t003

Download:

Table 4. The UEL distribution biases and MSEs under various estimation methods at δ = 2.5.

https://doi.org/10.1371/journal.pone.0288635.t004

The ML estimate (MLE), MPS estimate (MPSE), and Bayesian estimate of δ, λ and ϑ are calculated. Then, the biases and MSEs of the different estimates were determined as well as confidence intervals (CIs) were obtained. In MLE, the CIs by asymptotic CIs and bootstrap CIs with different algorithms as bootstrap-P (BP) and bootstrap-T (BT) were calculated. In MPS the confidence lengths (CLs) by asymptotic CIs are calculated. While in the Bayesian estimation method, the credible CIs (CCIs) by the highest posterior density CLs are obtained (see [32, 33]). Simulated results of bias, MSE, and length of CI (L.CI) were scheduled in Tables 3 and 4, and Fig 2.

Download:

Fig 2. Heatmaps of MSE values of different cases and different estimates: Left at δ = 0.5 & right at δ = 2.5.

https://doi.org/10.1371/journal.pone.0288635.g002

Some concluding remarks are noticed as the following:

As n increases, the bias and MSE across all estimates decrease.
The measures of MPSE are better than MLE.
MLEs and MPSEs are not as good as the metrics of Bayesian estimates.
The length of asymptotic CLs for MPSE has smaller values than the length of asymptotic CLs for MLEs.
The length of credible CLs for Bayesian has smaller values than the length of asymptotic CLs for MLE and MPSE.
The length of bootstrap CLs gets the smallest value of the length of CIs.
The length of bootstrap-T CLs gets the smallest value of the length of bootstrap-P CIs.

6 The UEL quantile regression model

According to the QF of the UEL distribution in Eq (11), we shall re-parametrize (5) in terms of the μ^th quantile μ = y(q; δ, λ, ϑ) such that λ can be written as follows: The PDF and CDF of the UEL distribution can be stated as follows with this parametrzation (17) where δ, ϑ > 0, 0 < μ < 1, 0 < y < 1 and 0 < q < 1, and the CDF is The notation Y ∼ UEL(δ, ϑ, μ) will be used from now on, where μ ∈ (0, 1) is the quantile parameter, δ, and ϑ are the shape parameters, and q ∈ (0, 1) is presumed to be known. The behavior of (17) can be investigated using the logarithmic scale for various values of δ, ϑ, μ and q. where , cannot be solved analytically in y. We can create a quantile regression model using the re-parametrized PDF (17), as in Ref. [34], where they used the Kumaraswamy distribution and unit Weibull regression model.

Let Y₁,…,Y_n be n independent random variables, where each Y_i, i=1,… n follows the PDF (17) with μ_i as unknown quantile parameter, and δ, ϑ (unknown shape parameter), where q ∈ (0, 1) is assumed to be known, i.e, Y_i ∼ UEL(δ, ϑ, μ_i). The quantile μ_i of Y_i must satisfy the following functional relation in order for the UEL quantile regression model to be defined here: where X_i = (1, X_1i, …, X_(p−1)i) represents the observations on p known covariates and B = (B₀, B₁, …, B_p−1)^T is a p-dimensional vector of unknown regression coefficients, p < n. We will suppose that the quantile link function g(.) maps (0, 1) into R and is strictly growing and twice differentiable. There are several choices for the link function g(.). For instance, the most popular link functions are:

logit: ;
probit: g (μ_i) = Φ⁻¹ (μ_i); where Φ⁻¹ (.) is the standard normal QF;
complementary log-log: g (μ_i) = ln [−ln (1 − μ_i)].

We only consider the logit as a link function in this paper because the parameters are directly translated into odds. Reference [35] gave their interpretation for logit when is the beta distribution’s mean. Then, we can write μ_i under the logit link function as follows: For assigned q ∈ (0, 1), let Θ = (B^T, δ, ϑ) be the vector of p unknown parameters to be estimated using the approach of ML. Using the structure of the PDF (17), the log- LF is given by: The MLE of parameters cannot be computed analytically and must be calculated numerically using an optimization algorithm such as Newton–Raphson or quasi-Newton. Under regularity conditions and when n is large, the asymptotic distribution of the MLE is approximately multivariate normal with mean vector (B^T, δ, ϑ) and V⁻¹ (Θ) is the variance-covariance matrix, where is the expected Fisher information matrix.

7 Data applications

This section contains three real-world data sets that demonstrate the UEL distribution’s modelling capabilities. Quantile regression modelling is addressed in the first and second data sets, while data modelling is addressed in the third data. Unit-Weibull regression (UWR), beta regression (BetaR) model, original linear regression (OLR) and quantile regression (QR) are three competitor regression models that are compared to the UEL quantile regression model for the first and second data. The QR has been obtained by “rq” function in “quantreg” package in the R program (see [36])). To compare between the UEL and the considered regression models, we compute Akaike’s information criterion (AIC), and Bayesian information criterion (BIC). For the third data set, the fits of the UEL distribution are compared with some other competitive models to illustrate the potentiality of the UEL model. The MLEs and their standard errors (SEs) are computed. The AIC, BIC, corrected AIC (CAIC), Hannan and Quinn information criteria (HQIC), Cramér–von Mises criterion (CVMC), Anderson-Darling criterion (ADC), and Kolmogorov-Smirnov distance (KSD) statistics, as well as their related P-values, were chosen as criteria measures for the third data set.

7.1 Quantile regression modeling for confidence of mock jurors in their verdicts

Data with responses of naive mock jurors to the conventional two-option verdict (guilt vs. acquittal) versus a three-option verdict setup (the third option was the Scottish ‘not proven’ alternative), in the presence/absence of conflicting testimonial evidence. A data frame containing 104 observations on three variables. The source of this data is Ref. [37] The following covariates are connected to this response variable Verdict: x₁ is a factor indicating whether a two-option or three-option verdict is requested, where two-option is -1, and three-option is 1. Conflict: x₂ is a factor. Is there conflicting testimonial evidence? If no is 1, yes is -1. The regression framework assumed for μ_i is provided by Logit μ_i = B₀ + B₁x_1i + B₂x_2i, i = 1, …, 104, where μ_i denotes the median q = 0.5 in the UEL regression (UELR) model. Table 5 gives the MLEs and SEs along with the AIC and BIC measures for the UELR, UWR, BetaR, OLR, and QR models.

Download:

Table 5. The MLE and SE with AIC and BIC for different models for mock jurors data.

https://doi.org/10.1371/journal.pone.0288635.t005

7.2 Quantile regression modeling for proportion of household income spent on food

Here, the proposed data on the proportion of income spent on food for a random sample of 38 households in a large US city. The source of this data is Ref. [38]. The covariates connected to this response variable are Income: x₁ is household income Persons: x₂ is the number of persons living in the household. The dependent variable is I(food/income). The regression pattern assumed for μ_i is as below Logit μ_i = B₀ + B₁x_1i + B₂x_2i, i = 1, …, 38 where μ_i denotes the median q = 0.5 in the UELR model. Table 6 gives the MLEs and SEs along with the AIC and BIC measures for the UELR, UWR, BetaR, OLR, and QR models. We note that UELR has a smaller value of AIC, and BIC than UWR, BetaR, OLR, and QR models, then the UELR model is better than UWR, BetaR, OLR, and QR models.

Download:

Table 6. The MLE and SE with AIC and BIC for different models for food expenditure data.

https://doi.org/10.1371/journal.pone.0288635.t006

Fig 3 provides the box plots for Mock Jurors data and food Expenditure data. Fig 4 discussed the profile likelihood function for food expenditure data which shows the likelihood estimators have maximum likelihood value.

Download:

Fig 3. Boxplot for dependent variables.

https://doi.org/10.1371/journal.pone.0288635.g003

Download:

Fig 4. Profile likelihood function for food expenditure data.

https://doi.org/10.1371/journal.pone.0288635.g004

7.3 Univariate modeling for Covid-19 data

Here, we will show how the UEL distribution may be used with Covid-19 data. Many other models are compared to the UEL distribution, including Kumaraswamy (Kom), Beta, Kumaraswamy Kumaraswamy (K-K) [38], Topp-Leone generalized exponential (TLGE) [39], Marshall-Olkin Kumaraswamy (MOK) [40], Topp-Leone Weibull Lomax (TLWL) [41], and type II power Topp-Leone inverse exponential (TIIPTLIE) [42]. The data contain 38 days of Covid-19 data from Saudi Arabia, from 22 July 2021 to 28 August 2021 (https://covid19.who.int). These results illustrate the drought mortality rate. The following data are listed in Table 7.

Download:

Table 7. COVID-19 data for Saudi Arabia.

https://doi.org/10.1371/journal.pone.0288635.t007

The MLEs and their SEs for the investigated models are provided in Table 8.

Download:

Table 8. MLE for distinct models with SE.

https://doi.org/10.1371/journal.pone.0288635.t008

For all fitted models using the Covid-19 data set, Table 9 shows the values of the AIC, BIC, CAIC, HQIC, CVMC, ADC, and KSD statistics, as well as their related P-values. Table 9 shows that the UEL distribution has the largest negative AIC, BIC, CAIC, HQIC, and P-value, as well as the smallest KSD, CVMC, and ADC values when compared to the other models used to fit the Covid-19. The empirical, histogram and PP-plot for the UEL distribution for Saudi Arabia’s Covid-19 data are shown in Fig 5.

Download:

Fig 5. The empirical CDF, histogram, and P-P plot for the UEL distribution for Covid-19 data of Saudi Arabia.

https://doi.org/10.1371/journal.pone.0288635.g005

Download:

Table 9. Various metrics for comparing various models.

https://doi.org/10.1371/journal.pone.0288635.t009

8 Summary and conclusion

This study proposes the unit-exponentiated Lomax distribution, based on an appropriate transformation, which is useful for modeling data on the unit interval. Some mathematical characteristics of the UEL distribution are explored, such as moments, PWMs, IM, entropy measures, and stress-strength reliability. The maximum likelihood, Bayesian, and maximum product of spacing methods are employed for estimating the parameter of the suggested distribution. Results from simulations show that the criteria measurements of Bayesian estimates are preferred to comparable alternative estimates. It can be demonstrated that the UEL regression model is a reasonable alternative to unit-Weibull regression, beta regression, and the original linear regression models when using mock jurors and food spending data. Using Covid-19 data, the proposed model outperforms the beta, Kumaraswamy-Kumaraswamy, Topp-Leone generalized exponential, Marshall-Olkin Kumaraswamy, Topp-Leone Weibull Lomax, and type II power Topp-Leone inverse exponential across a variety of comparison criteria. In future research, we will discuss the application of UEL distribution based on these points as [43–51].

References

1. Lomax K.S. Business failures: Another example of the analysis of failure data. Journal of American Statistical Association, 1954, 49, 847–852.
- View Article
- Google Scholar
2. Atkinson A.B.; Harrison A.J. Distribution of Personal Wealth in Britain. Cambridge University Press, Cambridge, 1978
3. Harris C.M. The Pareto distribution as a queue service discipline. Operations Research, 1968, 16, 307–313
- View Article
- Google Scholar
4. Corbellini, A., Crosato, L., Ganugi, P and Mazzoli, M. Fitting Pareto II distributions on firm size: Statistical methodology and economic puzzles. Paper presented at the International Conference on Applied Stochastic Models and Data Analysis, Chania, Crete, 2007
5. Holland O., Golaup A., Aghvami A.H. Traffic characteristics of aggregated module downloads for mobile terminal reconfiguration. IEE Proceedings on Communications, 2006, 135, 683–690
- View Article
- Google Scholar
6. Chen J., Addie R.G., and Zukerman M., Neame T.D. Performance evaluation of a queue fed by a Poisson Lomax Burst Process. IEEE Communication Letters, 2015, 19, 367–370.
- View Article
- Google Scholar
7. Campbell G. and Ratnaparkhi M.V. An application of Lomax distributions in receiver operating characteristic (ROC) curve analysis, Communications in Statistics-Theory Methods, 1993,22, 1681–1697.
- View Article
- Google Scholar
8. Arnold, B.C. Pareto Distributions, International Cooperative Publishing House, Fairland, Maryland, 1983
9. Vidondo B., Prairie Y.T., Blanco J.M. and Duarte C.M. Some aspects of the analysis of size spectra in aquatic ecology. Limnology and Oceanography 1997, 42, 184–192.
- View Article
- Google Scholar
10. Balkema A.A. and de Hann L. Residual life at great age. Annals of Probability, 1974, 972–804.
- View Article
- Google Scholar
11. Chahkandi M. and Ganjali M. On some lifetime distributions with decreasing failure rate. Computational Statistics and Data Analysis, 2009, 53, 4433–4440.
- View Article
- Google Scholar
12. Bryson M.C. Heavy-tailed distribution: properties and tests. Technometrics, 1974, 16, 161–168.
- View Article
- Google Scholar
13. Tadikamalla P.R. A look at the Burr and related distributions. International Statistical Review, 1980, 48, 337–344.
- View Article
- Google Scholar
14. Abdul-Moniem I.B.; Abdel-Hameed H.F. On exponentiated Lomax distribution. International Journal of Mathematical Archive, 2012, 3, 2144–2150.
- View Article
- Google Scholar
15. Ghitany M. E., Al-Awadhi F.A., and Alkhalfan L.A. Marshall Olkin extended Lomax distribution and its application to censored data. Communication in Statistics-Theory & Methods, 2007, 36, 1855–1866
- View Article
- Google Scholar
16. Lemonte A.J., and Cordeiro G.M. An extended Lomax distribution. Statistics, 2013, 47, 800–816.
- View Article
- Google Scholar
17. Tahir M.H., Cordeiro G.M., Mansoor M., and Zubair M. The Weibull–Lomax distribution: properties and applications. Hacettepe Journal of Mathematics and Statistics, 2015, 44, 461–480.
- View Article
- Google Scholar
18. Cordeiro G. M., Ortega E.M.M. and Popović B.V. The gamma-Lomax distribution. Journal of Statistical Computation and Simulation, 2015, 85, 305–319.
- View Article
- Google Scholar
19. Hassan A.S. and Abd-Alla M. Exponentiated Weibull-Lomax distribution: properties and estimation. Journal of Data Sciences, 2018, 16, 275–298.
- View Article
- Google Scholar
20. Abiodun A.A., and Ishaq A.I. On Maxwell-Lomax distribution: Properties and applications. Arab Journal of Basic and Applied Sciences, 2022, 29, 221–232.
- View Article
- Google Scholar
21. Nagarjuna V.B.V., Vardhan R.V., and Chesneau C. Nadarajah-Haghighi Lomax distribution and its applications. Mathematical and Computational Applications, 2022, 27, 1–13.
- View Article
- Google Scholar
22. Gómez-Déniz E., Sordo M.A., and Calderín-Ojeda E. The log-Lindley distribution as an alternative to the beta regression model with applications in insurance. Insurance: Mathematics and Economics, 2014, 54, 49–57
- View Article
- Google Scholar
23. Mazucheli J., Menezes A.F.B., and Dey S. The unit-Birnbaum–Saunders distribution with applications. Chilean Journal of Statistics, 2018, 9, 47–57.
- View Article
- Google Scholar
24. Mazucheli J., Menezes A.F., and Dey S. Unit-Gompertz distribution with applications. Statistica, 2019, 79, 25–43.
- View Article
- Google Scholar
25. Ghitany M., Mazucheli J., Menezes A., and Alqallaf F. The unit-inverse Gaussian distribution: A new alternative to two-parameter distributions on the unit interval. Communication in Statistics -Theory Methods, 2019, 48, 3423–3438.
- View Article
- Google Scholar
26. Mazucheli J., Menezes A.F.B., Fernandes L.B., De Oliveira R.P., and Ghitany M.E. The unit-Weibull distribution as an alternative to the Kumaraswamy distribution for the modeling of quantiles conditional on covariates. Journal of Applied Statistics, 2020, 47, 954–974. pmid:35706917
- View Article
- PubMed/NCBI
- Google Scholar
27. Korkmaz M.C., and Chesneau C. On the unit Burr-XII distribution with the quantile regression modeling and applications. Computational and Applied Mathematics, 2021, 40, 29.
- View Article
- Google Scholar
28. Bantan R.A.R., Jamal F., Chesneau C., and Elgarhy M. Theory and applications of the unit Gamma/Gompertz distribution. Mathematics, 2021, 9, 1850.
- View Article
- Google Scholar
29. Hassan A.S., Fayomi A., Algarni A., and Almetwally E.M. Bayesian and non-Bayesian inference for unit-exponentiated half-logistic distribution with data analysis. Applied Sciences, 2022, 12, 11253.
- View Article
- Google Scholar
30. Fayomi A., Hassan A.S., Baaqeel H.M., and Almetwally E.M. (2023). Bayesian inference and data analysis of the unit–power Burr X distribution. Axioms, 12, 297.
- View Article
- Google Scholar
31. Hassan A.S., and Alharbi R. S. Different estimation methods for the unit inverse exponentiated Weibull distribution. Communications for Statistical Applications and Methods, 2023, 30, 191–213
- View Article
- Google Scholar
32. Chen M.H., and Shao Q.M. Monte Carlo estimation of Bayesian credible and HPD intervals. Journal of Computational and Graphical Statistics, 1999, 8, 69–92.
- View Article
- Google Scholar
33. Mitnik, P.A., and Baek, S. The Kumaraswamy distribution: median-dispersion re-parameterizations for regression modeling and simulation-based estimation. Statistical Paper, 2013, 54, 177–192.
34. Ferrari S., and Cribari-Neto F. Beta regression for modelling rates and proportions. Journal of applied statistics, 2004, 31, 799–815.
- View Article
- Google Scholar
35. Konker, R. Quantile regression in R: A vignette. Version of April, 2017, 17, 88.
36. Smithson M., and Verkuilen J. A better Lemon Squeezer? Maximum-likelihood regression with beta-distributed dependent variables. Psychological Methods, 2006, 11, 54–71. pmid:16594767
- View Article
- PubMed/NCBI
- Google Scholar
37. Cribari-Neto F., and Zeileis A. Beta regression in R. Journal of Statistical Software, 2010, 34, 1–24.
- View Article
- Google Scholar
38. El-Sherpieny E.S.A., and Ahmed M.A. On the Kumaraswamy- Kumaraswamy distribution. International Journal of Basic and Applied Sciences, 2014, 3, 372–381.
- View Article
- Google Scholar
39. Sangsanit Y., and Bodhisuwan W. The Topp-Leone generator of distributions: properties and inferences. Songklanakarin Journal of Science and Technology, 2016, 38, 537–548.
- View Article
- Google Scholar
40. George R., and Thobias S. Marshall-Olkin Kumaraswamy distribution. International Mathematical Forum, 2017, 12, 47–69.
- View Article
- Google Scholar
41. Jamal, F., Reyad, H.M., Nasir, M. A., Chesneau, C., Shah, M.A.A., and Ahmed, S.O. Topp-Leone Weibull-Lomax distribution: properties, regression model and applications. HAL Id: hal-02270561, 2019, https://hal.archives-ouvertes.fr/hal-02270561
42. Bantan R.A., Jamal F., Chesneau C., and Elgarhy M. Type II power Topp-Leone generated family of distributions with statistical inference and applications. Symmetry, 2020, 12, 75.
- View Article
- Google Scholar
43. Abbas M., Ramli N., Majid A. A., & Ali J. M. (2014). The representation of circular arc by using rational cubic timmer curve. Mathematical Problems in Engineering, 2014.
- View Article
- Google Scholar
44. Maqsood S., Abbas M., Hu G., Ramli A. L. A., & Miura K. T. (2020). A novel generalization of trigonometric Bézier curve and surface with shape parameters and its applications. Mathematical Problems in Engineering, 2020, 1–25.
- View Article
- Google Scholar
45. Majeed A., Abbas M., Miura K. T., Kamran M., & Nazir T. (2020). Surface modeling from 2D contours with an application to craniofacial fracture construction. Mathematics, 8(8), 1246.
- View Article
- Google Scholar
46. Bashir U., Abbas M., & Ali J. M. (2013). The G2 and C2 rational quadratic trigonometric Bézier curve with two shape parameters with applications. Applied Mathematics and Computation, 219(20), 10183–10197.
- View Article
- Google Scholar
47. BiBi S., Abbas M., Misro M. Y., Hu G. (2019). A novel approach of hybrid trigonometric Bézier curve to the modeling of symmetric revolutionary curves and symmetric rotation surfaces. IEEE Access, 7, 165779–165792.
- View Article
- Google Scholar
48. Usman M., Abbas M., Miura K. T. (2020). Some engineering applications of new trigonometric cubic Bézier-like curves to free-form complex curve modeling. Journal of Advanced Mechanical Design, Systems, and Manufacturing, 14(4), JAMDSM0048–JAMDSM0048.
- View Article
- Google Scholar
49. Maqsood S., Abbas M., Miura K. T., Majeed A., Iqbal A. (2020). Geometric modeling and applications of generalized blended trigonometric Bézier curves with shape parameters. Advances in difference equations, 2020, 1–18.
- View Article
- Google Scholar
50. Abbas M., Majid A. A., Ali J. M. (2014). Positivity-preserving rational bi-cubic spline interpolation for 3D positive data. Applied Mathematics and Computation, 234, 460–476.
- View Article
- Google Scholar
51. BiBi S., Abbas M., Miura K. T., Misro M. Y. (2020). Geometric modeling of novel generalized hybrid trigonometric Bézier-like curve with shape parameters and its applications. Mathematics, 8(6), 967.
- View Article
- Google Scholar

[ref1] 1. Lomax K.S. Business failures: Another example of the analysis of failure data. Journal of American Statistical Association, 1954, 49, 847–852.
View Article
Google Scholar

[2] View Article

[3] Google Scholar

[ref2] 2. Atkinson A.B.; Harrison A.J. Distribution of Personal Wealth in Britain. Cambridge University Press, Cambridge, 1978

[ref3] 3. Harris C.M. The Pareto distribution as a queue service discipline. Operations Research, 1968, 16, 307–313
View Article
Google Scholar

[6] View Article

[7] Google Scholar

[ref4] 4. Corbellini, A., Crosato, L., Ganugi, P and Mazzoli, M. Fitting Pareto II distributions on firm size: Statistical methodology and economic puzzles. Paper presented at the International Conference on Applied Stochastic Models and Data Analysis, Chania, Crete, 2007

[ref5] 5. Holland O., Golaup A., Aghvami A.H. Traffic characteristics of aggregated module downloads for mobile terminal reconfiguration. IEE Proceedings on Communications, 2006, 135, 683–690
View Article
Google Scholar

[10] View Article

[11] Google Scholar

[ref6] 6. Chen J., Addie R.G., and Zukerman M., Neame T.D. Performance evaluation of a queue fed by a Poisson Lomax Burst Process. IEEE Communication Letters, 2015, 19, 367–370.
View Article
Google Scholar

[13] View Article

[14] Google Scholar

[ref7] 7. Campbell G. and Ratnaparkhi M.V. An application of Lomax distributions in receiver operating characteristic (ROC) curve analysis, Communications in Statistics-Theory Methods, 1993,22, 1681–1697.
View Article
Google Scholar

[16] View Article

[17] Google Scholar

[ref8] 8. Arnold, B.C. Pareto Distributions, International Cooperative Publishing House, Fairland, Maryland, 1983

[ref9] 9. Vidondo B., Prairie Y.T., Blanco J.M. and Duarte C.M. Some aspects of the analysis of size spectra in aquatic ecology. Limnology and Oceanography 1997, 42, 184–192.
View Article
Google Scholar

[20] View Article

[21] Google Scholar

[ref10] 10. Balkema A.A. and de Hann L. Residual life at great age. Annals of Probability, 1974, 972–804.
View Article
Google Scholar

[23] View Article

[24] Google Scholar

[ref11] 11. Chahkandi M. and Ganjali M. On some lifetime distributions with decreasing failure rate. Computational Statistics and Data Analysis, 2009, 53, 4433–4440.
View Article
Google Scholar

[26] View Article

[27] Google Scholar

[ref12] 12. Bryson M.C. Heavy-tailed distribution: properties and tests. Technometrics, 1974, 16, 161–168.
View Article
Google Scholar

[29] View Article

[30] Google Scholar

[ref13] 13. Tadikamalla P.R. A look at the Burr and related distributions. International Statistical Review, 1980, 48, 337–344.
View Article
Google Scholar

[32] View Article

[33] Google Scholar

[ref14] 14. Abdul-Moniem I.B.; Abdel-Hameed H.F. On exponentiated Lomax distribution. International Journal of Mathematical Archive, 2012, 3, 2144–2150.
View Article
Google Scholar

[35] View Article

[36] Google Scholar

[ref15] 15. Ghitany M. E., Al-Awadhi F.A., and Alkhalfan L.A. Marshall Olkin extended Lomax distribution and its application to censored data. Communication in Statistics-Theory & Methods, 2007, 36, 1855–1866
View Article
Google Scholar

[38] View Article

[39] Google Scholar

[ref16] 16. Lemonte A.J., and Cordeiro G.M. An extended Lomax distribution. Statistics, 2013, 47, 800–816.
View Article
Google Scholar

[41] View Article

[42] Google Scholar

[ref17] 17. Tahir M.H., Cordeiro G.M., Mansoor M., and Zubair M. The Weibull–Lomax distribution: properties and applications. Hacettepe Journal of Mathematics and Statistics, 2015, 44, 461–480.
View Article
Google Scholar

[44] View Article

[45] Google Scholar

[ref18] 18. Cordeiro G. M., Ortega E.M.M. and Popović B.V. The gamma-Lomax distribution. Journal of Statistical Computation and Simulation, 2015, 85, 305–319.
View Article
Google Scholar

[47] View Article

[48] Google Scholar

[ref19] 19. Hassan A.S. and Abd-Alla M. Exponentiated Weibull-Lomax distribution: properties and estimation. Journal of Data Sciences, 2018, 16, 275–298.
View Article
Google Scholar

[50] View Article

[51] Google Scholar

[ref20] 20. Abiodun A.A., and Ishaq A.I. On Maxwell-Lomax distribution: Properties and applications. Arab Journal of Basic and Applied Sciences, 2022, 29, 221–232.
View Article
Google Scholar

[53] View Article

[54] Google Scholar

[ref21] 21. Nagarjuna V.B.V., Vardhan R.V., and Chesneau C. Nadarajah-Haghighi Lomax distribution and its applications. Mathematical and Computational Applications, 2022, 27, 1–13.
View Article
Google Scholar

[56] View Article

[57] Google Scholar

[ref22] 22. Gómez-Déniz E., Sordo M.A., and Calderín-Ojeda E. The log-Lindley distribution as an alternative to the beta regression model with applications in insurance. Insurance: Mathematics and Economics, 2014, 54, 49–57
View Article
Google Scholar

[59] View Article

[60] Google Scholar

[ref23] 23. Mazucheli J., Menezes A.F.B., and Dey S. The unit-Birnbaum–Saunders distribution with applications. Chilean Journal of Statistics, 2018, 9, 47–57.
View Article
Google Scholar

[62] View Article

[63] Google Scholar

[ref24] 24. Mazucheli J., Menezes A.F., and Dey S. Unit-Gompertz distribution with applications. Statistica, 2019, 79, 25–43.
View Article
Google Scholar

[65] View Article

[66] Google Scholar

[ref25] 25. Ghitany M., Mazucheli J., Menezes A., and Alqallaf F. The unit-inverse Gaussian distribution: A new alternative to two-parameter distributions on the unit interval. Communication in Statistics -Theory Methods, 2019, 48, 3423–3438.
View Article
Google Scholar

[68] View Article

[69] Google Scholar

[ref26] 26. Mazucheli J., Menezes A.F.B., Fernandes L.B., De Oliveira R.P., and Ghitany M.E. The unit-Weibull distribution as an alternative to the Kumaraswamy distribution for the modeling of quantiles conditional on covariates. Journal of Applied Statistics, 2020, 47, 954–974. pmid:35706917
View Article
PubMed/NCBI
Google Scholar

[71] View Article

[72] PubMed/NCBI

[73] Google Scholar

[ref27] 27. Korkmaz M.C., and Chesneau C. On the unit Burr-XII distribution with the quantile regression modeling and applications. Computational and Applied Mathematics, 2021, 40, 29.
View Article
Google Scholar

[75] View Article

[76] Google Scholar

[ref28] 28. Bantan R.A.R., Jamal F., Chesneau C., and Elgarhy M. Theory and applications of the unit Gamma/Gompertz distribution. Mathematics, 2021, 9, 1850.
View Article
Google Scholar

[78] View Article

[79] Google Scholar

[ref29] 29. Hassan A.S., Fayomi A., Algarni A., and Almetwally E.M. Bayesian and non-Bayesian inference for unit-exponentiated half-logistic distribution with data analysis. Applied Sciences, 2022, 12, 11253.
View Article
Google Scholar

[81] View Article

[82] Google Scholar

[ref30] 30. Fayomi A., Hassan A.S., Baaqeel H.M., and Almetwally E.M. (2023). Bayesian inference and data analysis of the unit–power Burr X distribution. Axioms, 12, 297.
View Article
Google Scholar

[84] View Article

[85] Google Scholar

[ref31] 31. Hassan A.S., and Alharbi R. S. Different estimation methods for the unit inverse exponentiated Weibull distribution. Communications for Statistical Applications and Methods, 2023, 30, 191–213
View Article
Google Scholar

[87] View Article

[88] Google Scholar

[ref32] 32. Chen M.H., and Shao Q.M. Monte Carlo estimation of Bayesian credible and HPD intervals. Journal of Computational and Graphical Statistics, 1999, 8, 69–92.
View Article
Google Scholar

[90] View Article

[91] Google Scholar

[ref33] 33. Mitnik, P.A., and Baek, S. The Kumaraswamy distribution: median-dispersion re-parameterizations for regression modeling and simulation-based estimation. Statistical Paper, 2013, 54, 177–192.

[ref34] 34. Ferrari S., and Cribari-Neto F. Beta regression for modelling rates and proportions. Journal of applied statistics, 2004, 31, 799–815.
View Article
Google Scholar

[94] View Article

[95] Google Scholar

[ref35] 35. Konker, R. Quantile regression in R: A vignette. Version of April, 2017, 17, 88.

[ref36] 36. Smithson M., and Verkuilen J. A better Lemon Squeezer? Maximum-likelihood regression with beta-distributed dependent variables. Psychological Methods, 2006, 11, 54–71. pmid:16594767
View Article
PubMed/NCBI
Google Scholar

[98] View Article

[99] PubMed/NCBI

[100] Google Scholar

[ref37] 37. Cribari-Neto F., and Zeileis A. Beta regression in R. Journal of Statistical Software, 2010, 34, 1–24.
View Article
Google Scholar

[102] View Article

[103] Google Scholar

[ref38] 38. El-Sherpieny E.S.A., and Ahmed M.A. On the Kumaraswamy- Kumaraswamy distribution. International Journal of Basic and Applied Sciences, 2014, 3, 372–381.
View Article
Google Scholar

[105] View Article

[106] Google Scholar

[ref39] 39. Sangsanit Y., and Bodhisuwan W. The Topp-Leone generator of distributions: properties and inferences. Songklanakarin Journal of Science and Technology, 2016, 38, 537–548.
View Article
Google Scholar

[108] View Article

[109] Google Scholar

[ref40] 40. George R., and Thobias S. Marshall-Olkin Kumaraswamy distribution. International Mathematical Forum, 2017, 12, 47–69.
View Article
Google Scholar

[111] View Article

[112] Google Scholar

[ref41] 41. Jamal, F., Reyad, H.M., Nasir, M. A., Chesneau, C., Shah, M.A.A., and Ahmed, S.O. Topp-Leone Weibull-Lomax distribution: properties, regression model and applications. HAL Id: hal-02270561, 2019, https://hal.archives-ouvertes.fr/hal-02270561

[ref42] 42. Bantan R.A., Jamal F., Chesneau C., and Elgarhy M. Type II power Topp-Leone generated family of distributions with statistical inference and applications. Symmetry, 2020, 12, 75.
View Article
Google Scholar

[115] View Article

[116] Google Scholar

[ref43] 43. Abbas M., Ramli N., Majid A. A., & Ali J. M. (2014). The representation of circular arc by using rational cubic timmer curve. Mathematical Problems in Engineering, 2014.
View Article
Google Scholar

[118] View Article

[119] Google Scholar

[ref44] 44. Maqsood S., Abbas M., Hu G., Ramli A. L. A., & Miura K. T. (2020). A novel generalization of trigonometric Bézier curve and surface with shape parameters and its applications. Mathematical Problems in Engineering, 2020, 1–25.
View Article
Google Scholar

[121] View Article

[122] Google Scholar

[ref45] 45. Majeed A., Abbas M., Miura K. T., Kamran M., & Nazir T. (2020). Surface modeling from 2D contours with an application to craniofacial fracture construction. Mathematics, 8(8), 1246.
View Article
Google Scholar

[124] View Article

[125] Google Scholar

[ref46] 46. Bashir U., Abbas M., & Ali J. M. (2013). The G2 and C2 rational quadratic trigonometric Bézier curve with two shape parameters with applications. Applied Mathematics and Computation, 219(20), 10183–10197.
View Article
Google Scholar

[127] View Article

[128] Google Scholar

[ref47] 47. BiBi S., Abbas M., Misro M. Y., Hu G. (2019). A novel approach of hybrid trigonometric Bézier curve to the modeling of symmetric revolutionary curves and symmetric rotation surfaces. IEEE Access, 7, 165779–165792.
View Article
Google Scholar

[130] View Article

[131] Google Scholar

[ref48] 48. Usman M., Abbas M., Miura K. T. (2020). Some engineering applications of new trigonometric cubic Bézier-like curves to free-form complex curve modeling. Journal of Advanced Mechanical Design, Systems, and Manufacturing, 14(4), JAMDSM0048–JAMDSM0048.
View Article
Google Scholar

[133] View Article

[134] Google Scholar

[ref49] 49. Maqsood S., Abbas M., Miura K. T., Majeed A., Iqbal A. (2020). Geometric modeling and applications of generalized blended trigonometric Bézier curves with shape parameters. Advances in difference equations, 2020, 1–18.
View Article
Google Scholar

[136] View Article

[137] Google Scholar

[ref50] 50. Abbas M., Majid A. A., Ali J. M. (2014). Positivity-preserving rational bi-cubic spline interpolation for 3D positive data. Applied Mathematics and Computation, 234, 460–476.
View Article
Google Scholar

[139] View Article

[140] Google Scholar

[ref51] 51. BiBi S., Abbas M., Miura K. T., Misro M. Y. (2020). Geometric modeling of novel generalized hybrid trigonometric Bézier-like curve with shape parameters and its applications. Mathematics, 8(6), 967.
View Article
Google Scholar

[142] View Article

[143] Google Scholar

Figures

Abstract

1 Introduction

2 Model description

3 Characteristics of the UEL distribution

3.1 Probability weighted moments

3.2 Moments & associated measures

3.3 Quantile function

3.4 Incomplete moments

3.5 Information measures

3.6 Stress–strength model

4 Parameter estimation of the UEL model

4.1 ML method

4.2 MPS method

4.3 Bayesian method

5 Simulation study

6 The UEL quantile regression model

7 Data applications

7.1 Quantile regression modeling for confidence of mock jurors in their verdicts

7.2 Quantile regression modeling for proportion of household income spent on food

7.3 Univariate modeling for Covid-19 data

8 Summary and conclusion

References