Research article Special Issues

Generalized Bayesian inference study based on type-Ⅱ censored data from the class of exponential models

  • Generalized Bayesian (GB) is a Bayesian approach based on the learning rate parameter (LRP) (0<η<1) as a fraction of the power of the likelihood function. In this paper, we consider the GB method to perform inference studies for a class of exponential distributions. Generalized Bayesian estimators (GBE) and generalized empirical Bayesian estimators (GEBE) for the parameters of the considered distributions are obtained based on the censored type Ⅱ samples. In addition, generalized Bayesian prediction (GBP) and generalized empirical Bayesian prediction (GEBP) are considered using a one-sample prediction scheme. Monte Carlo simulations and illustrative example are performed for one parameter models to compare the performance of the GBE and GEBE estimation results and the GBP and GEBP prediction results for different values of the LRP.

    Citation: Yahia Abdel-Aty, Mohamed Kayid, Ghadah Alomani. Generalized Bayesian inference study based on type-Ⅱ censored data from the class of exponential models[J]. AIMS Mathematics, 2024, 9(11): 31868-31881. doi: 10.3934/math.20241531

    Related Papers:

    [1] Bing Long, Zaifu Jiang . Estimation and prediction for two-parameter Pareto distribution based on progressively double Type-II hybrid censored data. AIMS Mathematics, 2023, 8(7): 15332-15351. doi: 10.3934/math.2023784
    [2] Baria A. Helmy, Amal S. Hassan, Ahmed K. El-Kholy, Rashad A. R. Bantan, Mohammed Elgarhy . Analysis of information measures using generalized type-Ⅰ hybrid censored data. AIMS Mathematics, 2023, 8(9): 20283-20304. doi: 10.3934/math.20231034
    [3] Haidy A. Newer, Bader S Alanazi . Bayesian estimation and prediction for linear exponential models using ordered moving extremes ranked set sampling in medical data. AIMS Mathematics, 2025, 10(1): 1162-1182. doi: 10.3934/math.2025055
    [4] Haiping Ren, Xue Hu . Estimation for inverse Weibull distribution under progressive type-Ⅱ censoring scheme. AIMS Mathematics, 2023, 8(10): 22808-22829. doi: 10.3934/math.20231162
    [5] Heba S. Mohammed . Empirical E-Bayesian estimation for the parameter of Poisson distribution. AIMS Mathematics, 2021, 6(8): 8205-8220. doi: 10.3934/math.2021475
    [6] Ahmed R. El-Saeed, Ahmed T. Ramadan, Najwan Alsadat, Hanan Alohali, Ahlam H. Tolba . Analysis of progressive Type-Ⅱ censoring schemes for generalized power unit half-logistic geometric distribution. AIMS Mathematics, 2023, 8(12): 30846-30874. doi: 10.3934/math.20231577
    [7] Hassan Okasha, Mazen Nassar, Saeed A. Dobbah . E-Bayesian estimation of Burr Type XII model based on adaptive Type-Ⅱ progressive hybrid censored data. AIMS Mathematics, 2021, 6(4): 4173-4196. doi: 10.3934/math.2021247
    [8] Amal S. Hassan, Najwan Alsadat, Oluwafemi Samson Balogun, Baria A. Helmy . Bayesian and non-Bayesian estimation of some entropy measures for a Weibull distribution. AIMS Mathematics, 2024, 9(11): 32646-32673. doi: 10.3934/math.20241563
    [9] Magdy Nagy, Khalaf S. Sultan, Mahmoud H. Abu-Moussa . Analysis of the generalized progressive hybrid censoring from Burr Type-Ⅻ lifetime model. AIMS Mathematics, 2021, 6(9): 9675-9704. doi: 10.3934/math.2021564
    [10] Refah Alotaibi, Mazen Nassar, Zareen A. Khan, Ahmed Elshahhat . Analysis of Weibull progressively first-failure censored data with beta-binomial removals. AIMS Mathematics, 2024, 9(9): 24109-24142. doi: 10.3934/math.20241172
  • Generalized Bayesian (GB) is a Bayesian approach based on the learning rate parameter (LRP) (0<η<1) as a fraction of the power of the likelihood function. In this paper, we consider the GB method to perform inference studies for a class of exponential distributions. Generalized Bayesian estimators (GBE) and generalized empirical Bayesian estimators (GEBE) for the parameters of the considered distributions are obtained based on the censored type Ⅱ samples. In addition, generalized Bayesian prediction (GBP) and generalized empirical Bayesian prediction (GEBP) are considered using a one-sample prediction scheme. Monte Carlo simulations and illustrative example are performed for one parameter models to compare the performance of the GBE and GEBE estimation results and the GBP and GEBP prediction results for different values of the LRP.



    GB Generalized Bayesian
    LRP Learning rate parameter
    GBE Generalized Bayesian estimator
    GEBE Generalized empirical Bayesian estimator
    GBP Generalized Bayesian prediction
    GEBP Generalized empirical Bayesian prediction
    MLE Maximum likelihood estimator
    EB Empirical Bayes
    EBE Empirical Bayes estimator
    EBP Empirical Bayes prediction

    In the Bayesian inference techniques, GB analysis was introduced and studied based on the learning rate parameter (0<η<1). The traditional Bayesian framework for η=1 is a fraction of the power of the likelihood function L(θ)L(θ;data) for the parameter θΘ. This means that if the prior distribution of the parameter θ is π(θ), then the GB posterior distribution for θ is

    π(θ|data)Lη(θ)π(θ),θΘ,0<η<1. (1)

    In this study, the GBE, GEBE, GBP, and GEBP distributions from the class of exponential distributions are examined using type Ⅱ censored samples. Thus, the aim of this study is to examine all GB and GEB results for different LRP values, including η=1, which describe the traditional Bayes. For more information on the GB approach and how to select the value for the rate parameter, see [1,2,3,4,5,6,7,8,9,10,11,12,13]. Specifically, the choice of the learning rate was studied in [3,4,5,6] using the Safe Bayes algorithm based on the minimization of a sequential risk measure. In [7] and [8], another learning rate selection method was proposed, which included two different information adaptation strategies. The authors in [11] investigated GBE based on a joint type-Ⅱ censored sample from multiple exponential populations, using various values of the learning rate parameter. The same study was presented in [13] but was based on joint hybrid censoring. In [11,13] a range of values for the learning rate parameter have been chosen to obtain the best estimators for the parameters of the corresponding distributions, then GB results were compared with the traditional Bayesian results. Here we conduct our study based on different values of LRP to find out the effect of different values of LRP on the estimation and prediction results.

    A one-sample prediction scheme is a Bayesian prediction method that determines the point predictor or prediction interval for unknown future values in the same sample based on the currently available observations. A two-sample prediction scheme or a multiple-sample prediction scheme are two other ways in which Bayesian prediction can utilize currently available observations to predict one or more future samples. Numerous authors have addressed the prediction of future failures or samples using different censoring techniques in the context of different prediction methods. We highlight some points that are relevant to our research. For instance, [12] investigated the GBP using a combined type-Ⅱ censored sample drawn from multiple exponential populations. A study using a joint type-Ⅱ censored sample from two exponential populations for Bayes estimation and prediction was published in [14]. Based on a generalized order statistic and multiple type Ⅱ censoring, a Bayesian prediction for the future values of distributions from the class of exponential distributions was constructed in [15,16].

    In the Bayesian study, the parameter of the distribution under investigation is a random variable, i.e., this unknown parameter is distributed according to the prior distribution. Empirical Bayes (EB) is a Bayesian study in which the parameters of the prior distribution (hyperparameters) are also unknown. By combining the density function of the distribution and the prior distributions, we obtain the marginal density function of the hyperparameters, which is used to estimate the hyperparameters. Therefore, the data of the original distribution are used to find the maximum likelihood estimators (MLEs) of these hyperparameters. EB has been introduced by many authors; for example, [17] studied the empirical Bayes estimator (EBE) of reliability performances with progressive type-Ⅱ censoring of the Lomax model. The reliability and hazard function of the Kumaraswamy distribution were determined by [18] using progressive censored type Ⅱ samples to estimate the EBE of the parameters. The Rayleigh distribution was studied in [19] to determine EBE and empirical Bayes prediction (EBP).

    The rest of this article is organized as follows: Section 2 introduces the class of exponential models and then describes the problem of GB, GEB, GBP, and GEBP for this class. Section 3 applies the investigation from Section 2 to the exponential and Rayleigh models, which are given as examples of the one parameter exponential class. In Section 4, we present simulation study besides an illustrative example based on real data for the exponential and Rayleigh models to obtain the GBE, GEBE, GBP and GEBP for different LRP values and compare the results. Finally, Section 5 discusses the results and concludes the paper.

    In this section, we introduce the exponential class of models and examine the problems of the GB, GEB, GBP, and GEBP for this class.

    Let θ be the vector of parameters, define a function g(x;θ)g(x), and its derivative g(x) where limxg(x)=,limx0+g(x)=0. The probability density function (pdf), the cumulative probability density function (cdf) and the survival function (sf) of the exponential class are each given by:

    f(x;θ)=g(x)exp[g(x)],x>0,θ>0; (2)
    F(x;θ)=1exp[g(x)], (3)

    and

    ¯F(x;θ)=exp[g(x)]. (4)

    The likelihood function under type-Ⅱ censored data from the class is given by,

    L(x_;θ)=c¯F(xr;θ)nrri=1f(xi;θ)A(x_;θ)exp[B(x_;θ)], (5)

    where, c=n!(nr)!,A(x_;θ)=ri=1g(xi),B(x_;θ)=ri=1g(xi)+(nr)g(xr), x_=(x1,,xr).

    Consider the prior distribution of θ in the following general form:

    π(θ;δ)=Iδ1C(θ;δ)exp[D(θ;δ)], (6)

    where, Iδ=θC(θ;δ)exp[D(θ;δ)]dθ, δ is a vector of hyperparameters.

    Combining (5) and (6), after raising (5) to the fractional power η, the GB posterior distribution of θ is given by,

    πG(θ;δ,x_)=Iδ1Lη(x_;θ)π(θ;δ)=Iδ1G(θ;δ,x_)exp[H(θ;δ,x_)], (7)

    where Gδ=Aη(x_;θ)C(θ;δ),Hδ=ηB(x_;θ)+D(θ;δ), and Iδ=θGδexp[Hδ]dθ.

    Under the squared error loss function, then GBE is given by,

    ˆθGB=E(θ)=θθπG(θ;δ,x_)dθ. (8)

    Combining (1) and (6), to obtain the marginal pdf f(x;δ) as follows:

    f(x;δ)=θπ(θ;δ)f(x;θ)dθ=Iδ1θC(θ;δ)g(x)exp[{D(θ;δ)+g(x)}]dθ. (9)

    From pdf in (9) we obtain the cdf F(x;δ), then the likelihood function under type-Ⅱ censored data is given by,

    LE(x_;δ)=c¯F(xr;δ)nrri=1f(xi;δ). (10)

    Using the loglikelihood function LE(x_;δ)=logLE(x_;δ), to find the maximum likelihood estimator (MLE) ˆδ as follows:

    ˆδ=argmaxδLE(x_;δ). (11)

    By solving the following equation,

    LE(x_;δ)δ=0. (12)

    Substituting by ˆδ in (7), we obtain the posterior GE as follows:

    πGE(θ;ˆδ,x_)=Iˆδ1Gˆδexp[Hˆδ]. (13)

    Under the squared error loss function, GEBE is given by

    ˆθGE=E(θ)=θθπGE(θ;ˆδ,x_)dθ. (14)

    To determine the GBP and GEBP intervals using a one-sample prediction scheme under the type-Ⅱ censored sample from the class, the first r ordered statistics x_ are observed from a random sample of size n;r<n. A one-sample prediction scheme is considered to predict the rest of unobserved values xs,s=r+1,,n. The conditional density function of xs given x_ is given by

    f(xs|x_)=(nr)!(sr1)!(ns)![¯F(xr)¯F(xs)]sr1¯F(xs)ns¯F(xr)(nr)f(xs). (15)

    Substituting by (2), (4) in (15), the conditional density function of xs given x_ is,

    f(xs|x_)=sr1j=0cjg(xs)exp[nj{g(xs)g(xr)}], (16)

    where, cj=(1)srj1(nr)!j!(srj1)!(ns)!,nj=nrj.

    Combining (7) and (16), then integrating with respect to θ, the GB predictive density function is given by,

    fG(xs|x_)=Iδ1sr1j=0cjθg(xs)Gδexp[{nj(g(xs)g(xr))+Hδ}]dθ. (17)

    The predictive reliability function of xs,s=r+1,,n is given by:

    ¯Fδ(t|x_)=Iδ1sr1j=0cjnjθGδexp[{nj(g(t)g(xr))+Hδ}]dθ. (18)

    Equation (18) to 1+α2and1α2, respectively, we obtain (1α)% GBP bounds (Lδ,Uδ). GEBP bounds (Lˆδ,Uˆδ), can be obtained by substituting by ˆδ in (18) then equating to 1+α2and1α2.

    In this section, we apply the results in the previous section to one parameter models; therefore, the models that are discussed here are exponential Exp(θ) and Rayliegh Ray(θ). For these two distributions, the parameter θ assumed to be unknown, we may consider the conjugate prior distribution of θ as a gamma prior distribution, θGam(δ1,δ2), hence,

    C(θ;δ)=θδ11,D(θ;δ)=δ2θ;I1δ=δ2δ1Γ(δ1),δ1,δ2>0. (19)

    Here we give the essential functions and important forms derived in Section 2 and Eq (19) for the exponential model as follows:

    g(x)=θx,x>0. (20)

    For the likelihood function, we have

    A(x_;θ)=θr,TE=ri=1xi+(nr)xrandB(x_;θ)=θTE.

    Generalized posterior function can be formed from the following:

    Gδ=θηr+δ11,Hδ=θ(ηTE+δ2)andIδ=Γ(ηr+δ1)(ηTE+δ2)ηr+δ1.

    The GBE of the parameter θ is given by,

    ˆθGB=ηr+δ1ηTE+δ2. (21)

    The predictive reliability function of xs is given by,

    ¯Fδ(t|x_)=sr1j=0cjnj[1+nj(txr)ηTE+δ2](ηr+δ1). (22)

    Equating (22) to 1+α2and1α2, respectively, we obtain (1α)% GBP bounds (Lδ,Uδ).

    The functions and forms under GEB study can be illustrated as follows:

    The marginal pdf f(x;δ) is given in the following form,

    f(x;δ)=δ1δ2δ1(x+δ2)(δ1+1). (23)

    Using pdf f(x;δ) and cdf F(x;δ), we obtain the likelihood function based on type-Ⅱ censored data, as follows:

    LE(x_;δ)(δ1δ2δ1)rri=1(xi+δ2)(δ1+1)(1+xrδ2)(nr)δ1. (24)

    By differentiating the loglikelihood function LE(x_;δ) w. r. to δ1andδ2 and equating each equation to zero, then solving them numerically, we obtain the estimators ˆδ1andˆδ2.

    The GEBE of the parameter θ is given by,

    ˆθGE=ηr+ˆδ1ηTE+ˆδ2. (25)

    The predictive reliability function of xs is given by,

    ¯Fˆδ(t|x_)=sr1j=0cjnj[1+nj(txr)ηTE+ˆδ2](ηr+ˆδ1). (26)

    Equating (26) to (1+α)/2and(1α)/2, respectively, we obtain (1α)% GEBP bounds (Lˆδ,Uˆδ).

    The essential functions and important forms derived in Section 2 and Eq (19) for Rayliegh distribution are derived as follows:

    g(x)=θx22,x>0. (27)

    For the likelihood function we have

    A(x_;θ)=θrri=1xi,TR=ri=1x2i/2+(nr)x2r/2andB(x_;θ)=θTR.

    Generalized posterior function, can be formed from the following:

    Gδ=θηr+δ11(ri=1xi)η,Hδ=θ(ηTR+δ2)andIδ=(ri=1xi)ηΓ(ηr+δ1)(ηTR+δ2)ηr+δ1.

    The GBE of the parameter θ is given by,

    ˆθGB=ηr+δ1ηTR+δ2. (28)

    The predictive reliability function of xs is given by,

    ¯Fδ(t|x_)=sr1j=0cjnj[1+nj(t2x2r)/2ηTR+δ2](ηr+δ1). (29)

    Equating (29) to 1+α2and1α2, respectively, we obtain (1α)% GBP bounds (Lδ,Uδ).

    The marginal pdf f(x;δ) is given in the following form:

    f(x;δ)=δ1δ2δ1(x+δ2)(δ1+1). (30)

    Using pdf f(x;δ) and cdf F(x;δ) we obtain the likelihood function based on type-Ⅱ censored data, as follows:

    LE(x_;δ)(δ1δ2δ1)rri=1(xi+δ2)(δ1+1)(1+xr/δ2)(nr)δ1. (31)

    By differentiating the loglikelihood function LE(x_;δ) with respect to δ1andδ2 and equating each equation to zero, then solving them numerically, we obtain the estimators ˆδ1andˆδ2.

    The GEBE of the parameter θ is given by,

    ˆθGE=ηr+ˆδ1ηTE+ˆδ2. (32)

    The predictive reliability function of xs is given by,

    ¯Fˆδ(t|x_)=sr1j=0cjnj[1+nj(t2x2r)/2ηTR+ˆδ2](ηr+ˆδ1). (33)

    Equating (32) to 1+α2and1α2, respectively, we obtain (1α)% GEBP bounds (Lˆδ,Uˆδ).

    In this section, a simulation study for the exponential and Rayleigh distributions is presented to obtain the GBE, GEBE, GBP, and GEBP for different LRP values and compare the results. An example based on real data for exponential and Rayleigh distributions is given to illustrate the results.

    In this subsection, the results of the Monte Carlo simulation are presented to evaluate the performance of the inference methods derived in the previous sections. The simulation study is designed and carried out for the two models as follows:

    ● Generate one sample from each distribution with size n=50 and choosing r=30,40,50.

    ● Based on the chosen values of the hyperparameters (δ1,δ2)=(4,2), the suggested value for the parameter is θ=2, whereθ is obtained as the mean of gamma distribution in (18).

    ● For EB, we use MLE (ˆδ1,ˆδ2) to compute ˆθGE, where the results of MLE (ˆδ1,ˆδ2) based on exponential and Rayliegh distributions are shown in Table 1.

    Table 1.  The MLEs of the hyperparameters ˆδ1,ˆδ2 under different data from the two distributions.
    Exp(θ) Ray(θ)
    (n,r) (ˆδ1,ˆδ2)
    (20, 15) (10.497, 4.1) (10.95, 4.625)
    (50, 30) (11.31, 4.926) (11, 4.4252)
    (50, 40) (10.9, 5.218) (10.995, 5.1456)
    (50, 50) (10.3, 5.289) (10.5, 5.2727)

     | Show Table
    DownLoad: CSV

    ● For the Monte Carlo simulations, we use M=10,000 replicates; therefore, the estimator ˆθ=Mi=1ˆθiM, and the estimated risk, ER=Mi=1(θiˆθi)2M.

    ● Using (21), (25), (28) and (32), the estimation results are obtained and expressed by the estimator ˆθ and ER for different values of LRP, where η=0.1,0.5,and1.

    ● The results of GBE and GEBE for exponential and Rayliegh distributions are shown in Tables 2 and 4.

    Table 2.  GBE and GEBE for the parameter of exponential distribution.
    r η ˆθGB ERGB ˆθGE ERGE
    30
    40
    50
    0.1 2.0491 0.0033 2.2308 0.0070
    2.0436 0.0027 2.0687 0.0064
    2.0295 0.0023 1.9677 0.0067
    30
    40
    50
    0.5 2.0611 0.0078 2.1430 0.0058
    2.0461 0.0052 2.0526 0.0053
    2.0401 0.0046 2.0052 0.0047
    30
    40
    50
    1 2.0655 0.0127 2.1119 0.0051
    2.0495 0.0111 2.0495 0.0046
    2.0414 0.0078 2.0187 0.0035

     | Show Table
    DownLoad: CSV

    ● Prediction results are based on one sample from each distribution with size n=20, the number of observations is r=15. We then compute the GBP, GEBP bounds, and their lengths at α=0.05, for the future values with s=16,18,20 using (22), (26), (29), and (33).

    ● The results of GBP and GEBP for exponential and Rayliegh distributions are shown in Tables 3 and 5.

    Table 3.  GBP and GEBP bound for exponential future values.
    s η (L,U)GB length (L,U)GE length
    16
    18
    20
    0.1 (0.6596, 1.1826) 0.5230 (0.6591, 1.0062) 0.3471
    (0.7282, 2.1919) 1.4637 (0.7178, 1.6180) 0.9002
    (0.9338, 4.9415) 4.0077 (0.8992, 3.3181) 2.4189
    16
    18
    20
    0.5 (0.6597, 1.0920) 0.4323 (0.6592, 1.0142) 0.3550
    (0.7321, 1.8567) 1.1246 (0.7237, 1.6225) 0.8988
    (0.9557, 3.9800) 3.0243 (0.9251, 3.3245) 2.3994
    16
    18
    20
    1 (0.6596, 1.0642) 0.4046 (0.6593, 1.0181) 0.3588
    (0.7337, 1.7554) 1.0217 (0.7273, 1.6225) 0.8952
    (0.9653, 3.6914) 2.7261 (0.9411, 3.3220) 2.3809

     | Show Table
    DownLoad: CSV

    From Table 2. According to ˆθ and ER, GBE becomes better for small values of LRP but for large values of r, which means getting the best result at η=0.1 and r=50 (complete sample). GEBE becomes better for large values of LRP and for large values of r, which means getting the best result at η=1 and r=50. In general, the result of GBE is better than that of GEBE.

    From Table 3. According to the length of the interval, GBP and GEBP become better for large values of LRP, which means getting the best result at η=1 and s=16. In general, the result of GEBP is better than that of GBP.

    From Table 4. GBE becomes better for small values of LRP but for large values of r, which means getting the best result at η=0.1 and r=50 (complete sample). GEBE becomes better for large values of LRP and for large values of r, except for the complete sample, the result becomes better for small values of LRP which means getting the best result at η=0.1 and r=50. The result of GBE is better than that of GEBE at r=30,40, but GEBE is better than GBE for the complete sample.

    Table 4.  GBE and GEE for the parameter of Rayleigh distribution.
    r η ˆθGB ERGB ˆθGE ERGE
    30
    40
    50
    0.1 2.0131 0.0028 2.3676 0.0129
    2.0124 0.0023 2.1026 0.0082
    2.0122 0.0013 1.9992 0.0013
    30
    40
    50
    0.5 2.0415 0.0060 2.2088 0.0095
    2.0321 0.0026 2.0677 0.0048
    2.0311 0.0024 2.0166 0.0015
    30
    40
    50
    1 2.0514 0.0075 2.1532 0.0074
    2.0436 0.0059 2.0603 0.0020
    2.0349 0.0041 2.0260 0.0018

     | Show Table
    DownLoad: CSV

    From Table 5. According to the length of the interval, GBP and GEBP become better for large values of LRP, which means getting the best result at η=1 and s=16. In general, the result of GEBP is better than that of GBP.

    Table 5.  GBP and GEBP bound for Rayleigh future values.
    s η (L,U)GB length (L,U)GE length
    16
    18
    20
    0.1 (1.1379, 1.5297) 0.3918 (1.1376, 1.4252) 0.2876
    (1.1967, 2.0880) 0.8913 (1.1914, 1.8238) 0.6324
    (1.3577, 3.1400) 1.7823 (1.3448, 2.6322) 1.2874
    16
    18
    20
    0.5 (1.1379, 1.4695) 0.3316 (1.1377, 1.4253) 0.2876
    (1.2000, 1.9209) 0.7209 (1.1954, 1.8145) 0.6191
    (1.3738, 2.8175) 1.4437 (1.3597, 2.6121) 1.2524
    16
    18
    20
    1 (1.1379, 1.4507) 0.3128 (1.1378, 1.4251) 0.2873
    (1.2013, 1.8674) 0.6661 (1.976, 1.8084) 0.6108
    (1.3807, 2.7132) 1.3325 (1.3689, 2.5993) 1.2304

     | Show Table
    DownLoad: CSV

    To illustrate the results, two examples based on real data are given for exponential and Rayleigh distributions, respectively. For the GB study, there is no information about the prior, and noninformative prior should be used; therefore, we suggest the hyperparameters as (δ1,δ2)=(0.0001,0.0001), which results in the MLE for the parameter, which means there is no effect for the LRP in the case of noninformative prior.

    The data in Table 6, contains times to breakdown of an insulating fluid between electrodes recorded at 34kv (see, [20]). Table 7 provide the MLEs of ˆδ1,ˆδ2 under breakdown data from the two distributions.

    Table 6.  Breakdown time data (n=19).
    0.19, 0.78, 0.96, 1.31, 2.78, 3.16, 4.15, 4.67, 4.85, 6.50, 7.35, 8.01, 8.27, 12.06, 31.75, 32.52, 33.91, 36.71, 72.89

     | Show Table
    DownLoad: CSV
    Table 7.  The MLEs of ˆδ1,ˆδ2 under breakdown data from the two distributions.
    Exp(θ) Ray(θ)
    (n,r) (ˆδ1,ˆδ2)
    (19, 10) (7.427, 1.1) (7.211, 0.64)
    (19, 15) (6.658, 1) (5.767, 0.55)
    (19, 19) (5.794, 0.95) (4.685, 0.5)

     | Show Table
    DownLoad: CSV

    From Table 8, the results of GBE are nearly the MLE for the parameter in the case of noninformative prior, as we see there is no effect for the LRP in both distributions, and the best result for exponential distribution is ˆθGB=0.0696, while the best result for Rayliegh distribution is ˆθGB=0.0036. GEBE becomes better for large values of LRP and for large values of r, which means the best result for the exponential distribution is ˆθGE=0.0905, while the best result for the Rayliegh distribution is ˆθGE=0.0045. Because of using noninformative prior, the result of GEBE is better than that of GBE.

    Table 8.  GBE and GEBE for the parameters of the two distributions.
    Exp(θ) Ray(θ)
    r η ˆθGB ˆθGE ˆθGB ˆθGE
    10
    15
    19
    0.1 0.1138 0.8525 0.0395 0.31651
    0.0670 0.3489 0.0054 0.0263
    0.0696 0.2725 0.0036 0.0127
    10
    15
    19
    0.5 0.1138 0.2760 0.0395 0.0960
    0.0670 0.1254 0.0054 0.0096
    0.0696 0.1113 0.0036 0.0055
    10
    15
    19
    1 0.1138 0.1959 0.0395 0.0678
    0.0670 0.0963 0.0054 0.0075
    0.0696 0.0905 0.0036 0.0045

     | Show Table
    DownLoad: CSV

    From Tables 9 and 10, according to the length of the interval, GBP becomes better for large values of LRP, which means getting the best result at η=1 and s=16, but GEBP becomes better for small value of LRP, that means getting the best result at η=0.1 and s=16. In general, the result of GEBP is better than that of GBP.

    Table 9.  GBP and GEBP bound for exponential future values.
    s η (L,U)GB length (L,U)GE length
    16
    19
    0.1 (31.845, 91.575) 59.723 (31.768, 35.092) 3.324
    (36.975,488.5) 451.525 (33.067, 51.565) 18.498
    16
    19
    0.5 (31.845, 49.523) 17.678 (31.85, 40.151) 8.301
    (38.56,137.8) 99.240 (35.561, 80.011) 44.450
    16
    19
    1 (31.845, 47.349) 15.504 (31.816, 42.185) 10.369
    (38.9,121.15) 82.250 (36.807, 90.784) 53.977

     | Show Table
    DownLoad: CSV
    Table 10.  GBP and GEBP bound for Rayliegh future values.
    s η (L,U)GB length (L,U)GE length
    16
    19
    0.1 (31.787, 49.785) 17.998 (31.758, 33.154) 1.396
    (33.711,110.561) 76.85 (32.289, 39.415) 7.126
    16
    19
    0.5 (31.787, 38.009) 6.222 (31.771, 35.048) 3.277
    (34.285, 60.122) 25.837 (33.269, 47.731) 14.462
    16
    19
    1 (31.786, 37.30) 5.514 (31.776, 35.712) 3.936
    (34.406, 56.605) 22.199 (33.718, 50.222) 16.504

     | Show Table
    DownLoad: CSV

    In this study, one-parameter models belonging to the class of exponential models are considered. Two well-known models, Exp(θ) and Ray(θ), are examined based on a censored type-Ⅱ sample. GB, GEB, GBP, and GEBP are discussed for the two distributions with different values of LRP η. In the following subsections, we discuss simulation and illustrative example results.

    From the results in Table 2 to Table 5, we can summarize the results of the two distributions as follows:

    ● GBE becomes better for small values of LRP but for large values of r, which means getting the best result at η=0.1 and r=50. GEBE becomes better for large values of LRP and for large values of r, which means getting the best result at η=1 and r=50.

    ● GBP and GEBP become better for large values of rand LRP, which means getting the best result at η=1 and r=50.

    ● The result of GBE is better than that of GEBE, but the result of GEBP is better than that of GBP.

    ● Small values of LRP give the best result for GBE but vice versa for GEBP.

    ● GBE becomes better for small values of LRP but for large values of r, which means getting the best result at η=0.1 and r=50. GEBE becomes better for large values of LRP and for large values of r, except for the complete sample, the result becomes better for small values of LRP, which means getting the best result at η=0.1 and r=50. The result of GBE is better than that of GEBE at r=30,40, but GEBE is better than GBE for the complete sample.

    ● GBP and GEBP become better for large values of rand LRP, which means getting the best result at η=1 and r=50.

    ● The result of GBE is better than that of GEBE, but the result of GEBP is better than that of GBP.

    ● Small values of LRP for the complete sample give the best result for GBE but vice versa for GEBP.

    In this subsection, because of using noninformative prior, we can say that GBE is MLE. From the results in Table 6 to Table 10, we can summarize the results of estimation and prediction for the two distributions as follows:

    ● There is no effect for the LRP on the results of GBE in the case of noninformative prior.

    ● The best result of GBE for exponential distribution is ˆθGB=0.0696.

    ● The best result of GEBE for exponential distribution is ˆθGE=0.0905.

    ● The best result of GBE for Rayliegh distribution is ˆθGB=0.0036.

    ● The best result of GEBE for Rayliegh distribution is ˆθGB=0.0045.

    ● In both distributions, GEBE becomes better for large values of LRP and for large values of r, which means getting the best result for the complete sample at η=1.

    ● Because of using the noninformative prior, the result of GEBE is better than that of GBE (MLE).

    ● According to the length of the interval, GBP becomes better for large values of LRP, which means getting the best result at η=1 and s=16.

    ● GEBP becomes better for small values of LRP, which means getting the best result at η=0.1 and s=16.

    ● The result of GEBP is better than that of GBP (because of using noninformative prior).

    Generally, we can conclude that the result of GBE is better than that of GEBE, but the result of GEBP is better than GBP. Small values of LRP for the complete sample mostly give the best result for GBE and GEBE but LRP differ for the best result of GBP and GEBP.

    The study here is based on one-parameter models; in future work, the study can be extended to two or more parameters.

    Yahia Abdel-Aty: Project administration, methodology, investigation; Mohamed Kayid: Writing – original draft, formal analysis, data curation, conceptualization; Ghadah Alomani: Writing – review & editing, supervision, software, resources, funding acquisition. All authors have read and approved the final version of the manuscript for publication.

    Princess Nourah bint Abdulrahman University Researchers Supporting Project number (PNURSP2024R226), Princess Nourah bint Abdulrahman University, Riyadh, Saudi Arabia.

    There are no conflict of interest.



    [1] J. W. Miller, D. B. Dunson, Robust Bayesian inference via coarsening, J. Amer. Statist. Assoc., 114 (2019), 1113–1125. https://doi.org/10.1080/01621459.2018.1469995 doi: 10.1080/01621459.2018.1469995
    [2] P. Grünwald, The safe Bayesian: Learning the learning rate via the mixability gap, In: Algorithmic learning theory, Heidelberg: Springer, 7568 (2012), 169–183. https://doi.org/10.1007/978-3-642-34106-9_16
    [3] P. Grünwald, T. Van Ommen, Inconsistency of Bayesian inference for misspacified linear models, and a proposal for repairing it, Bayesian Anal., 12 (2017), 1069–1103. https://doi.org/10.1214/17-BA1085 doi: 10.1214/17-BA1085
    [4] P. Grünwald, Safe probability, J. Statist. Plann. Inference, 195 (2018), 47–63. https://doi.org/10.1016/j.jspi.2017.09.014 doi: 10.1016/j.jspi.2017.09.014
    [5] R. Heide, A. Kirichenko, P. Grünwald, N. Mehta, Safe-Bayesian generalized linear regression, In: Proceedings of the twenty third international conference on artificial intelligence and statistics, 108 (2020), 2623–2633.
    [6] C. C. Holmes, S. G. Walker, Assigning a value to a power likelihood in a general Bayesian model, Biometrika, 104 (2017), 497–503. https://doi.org/10.1093/biomet/asx010 doi: 10.1093/biomet/asx010
    [7] S. P. Lyddon, C. C. Holmes, S. G. Walker, General Bayesian updating and the loss-likelihood bootstrap, Biometrika, 106 (2019), 465–478. https://doi.org/10.1093/biomet/asz006 doi: 10.1093/biomet/asz006
    [8] R. Martin, Invited comment on the article by van der Pas, Szabó, and van der Vaart, Bayesian Anal., 12 (2017), 1254–1258.
    [9] R. Martin, B. Ning, Empirical priors and coverage of posterior credible sets in a sparse normal mean model, Sankhyā A, 82 (2020), 477–498. https://doi.org/10.1007/s13171-019-00189-w doi: 10.1007/s13171-019-00189-w
    [10] P. S. Wu, R. Martin, A comparison of learning rate selection methods in generalized Bayesian inference, Bayesian Anal., 18 (2023), 105–132. https://doi.org/10.1214/21-BA1302 doi: 10.1214/21-BA1302
    [11] Y. Abdel-Aty, M. Kayid, G. Alomani, Generalized Bayes estimation based on a joint type-Ⅱ censored sample from k-exponential populations, Mathematics, 11 (2023), 2190. https://doi.org/10.3390/math11092190 doi: 10.3390/math11092190
    [12] Y. Abdel-Aty, M. Kayid, G. Alomani, Generalized Bayes prediction study based on joint type-Ⅱ censoring, Axioms, 12 (2023), 716. https://doi.org/10.3390/axioms12070716 doi: 10.3390/axioms12070716
    [13] Y. Abdel-Aty, M. Kayid, G. Alomani, Selection effect of learning rate parameter on estimators of k exponential populations under the joint hybrid censoring, Heliyon, 10 (2024), e34087. https://doi.org/10.1016/j.heliyon.2024.e34087 doi: 10.1016/j.heliyon.2024.e34087
    [14] A. R. Shafay, N. Balakrishnan, Y. Abdel-Aty, Bayesian inference based on a jointly type-Ⅱ censored sample from two exponential populations, J. Stat. Comput. Simul., 84 (2014), 2427–2440. https://doi.org/10.1080/00949655.2013.813025 doi: 10.1080/00949655.2013.813025
    [15] Y. Abdel-Aty, J. Franz, M. A. W. Mahmoud, Bayesian prediction based on generalized order statistics using multiply type-Ⅱ censoring, Statistics, 41 (2007), 495–504. https://doi.org/10.1080/02331880701223357 doi: 10.1080/02331880701223357
    [16] A. R. Shafay, M. M. Mohie El-Din, Y. Abdel-Aty, Bayesian inference based on multiply type-Ⅱ censored sample from a general class of distributions, J. Stat. Theory Appl., 17 (2018), 146–157. https://doi.org/10.2991/jsta.2018.17.1.11 doi: 10.2991/jsta.2018.17.1.11
    [17] M. M. Mohie El-Din, H. Okasha, B. Al-Zahrani, Empirical Bayes estimators of reliability performances using progressive type-Ⅱ censoring from Lomax model, J. Adv. Res. Appl. Math., 5 (2013), 74–83.
    [18] M. Kumar, S. K. Singh, U. Singh, A. Pathak, Empirical Bayes estimator of parameter, reliability and hazard rate for Kumaraswamy distribution, Life Cycle Reliab. Saf. Eng., 8 (2019), 243–256. https://doi.org/10.1007/s41872-019-00085-0 doi: 10.1007/s41872-019-00085-0
    [19] M. Al-Ameen, Y. Abdel-Aty, Empirical Bayes inference for Rayleigh distribution, J. Stat. Appl. Pro., 11 (2022), 695–708. http://dx.doi.org/10.18576/jsap/110226 doi: 10.18576/jsap/110226
    [20] W. Nelson, Applied life data analysis, New York: Wiley, 1982. https://doi.org/10.1002/0471725234
  • Reader Comments
  • © 2024 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(528) PDF downloads(49) Cited by(0)

Figures and Tables

Tables(10)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog