Subscribe Now Subscribe Today
Research Article
 

Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data



Al Omari Mohammed Ahmed
 
Facebook Twitter Digg Reddit Linkedin StumbleUpon E-mail
ABSTRACT

Weibull distribution has drawn a lot of attention from statisticians working on both theory and methods and even other fields of applied statistics. In this study, the performance of Maximum Likelihood Estimator and the Bayesian Estimator using Jeffreys prior and extension of Jeffreys prior information for estimating the scale parameter, survival function and the hazard function of the Weibull distribution given shape with type-I and II censored data was estimated. The comparisons are made with respect to the Mean Square Error (MSE) for various sample sizes with different values of the scale and shape parameters, extension of Jeffreys prior and different percentage of censoring. The Bayesian using extension of Jeffreys prior based on type-I and II is better estimator compared to the other estimations.

Services
Related Articles in ASCI
Search in Google Scholar
View Citation
Report Citation

 
  How to cite this article:

Al Omari Mohammed Ahmed , 2014. Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data. Trends in Applied Sciences Research, 9: 438-449.

DOI: 10.3923/tasr.2014.438.449

URL: https://scialert.net/abstract/?doi=tasr.2014.438.449
 
Received: June 01, 2014; Accepted: September 02, 2014; Published: September 18, 2014



INTRODUCTION

The Weibull distribution has the widest variety of many applications in different areas, including lifetime testing, reliability theory and others. Sinha (1986) estimated the scale and shape parameters of Weibull distribution by MLE and Bayesian methods. Smith and Naylor (1987) developed the maximum likelihood and Bayesian approach and compared them using the Weibull distribution. Sun (1997) estimated the two parameters of Weibull distribution where he compared Jeffreys prior with that of the reference prior under Bayesian methods. Hossain and Zimmer (2003) estimated the parameters of Weibull distribution using complete and censored samples by MLE and least squares method. Singh et al. (2005) obtained Bayesian and MLE for Exponentiated Weibull distribution based type-II censoring. Soliman et al. (2006) estimated Weibull distribution by using MLE and Bayesian estimation following by estimated the hazard and reliability functions. Kantar and Senoglu (2008) reported their findings on the comparative study for the location and scale parameters of the Weibull distribution with a given shape parameter. Kundu and Howlader (2010) obtained Bayesian inference and prediction of the inverse Weibull distribution based type-II censored data. Pandey et al. (2011) compared Bayesian estimation and MLE of the scale parameter in Weibull distribution with known shape.

The objective of this study is to estimate the scale parameter, survival and hazard functions of the Weibull distribution based on type-I and II censored data by using Bayesian approach and compared to MLE using Mean Square Error (MSE) to determine the better estimator under several conditions.

MATERIALS AND METHODS

Maximum likelihood estimation of Weibull censored data: Concept of maximum likelihood estimation on Weibull distribution with censored data was introduced here. Let (x1,…, xn) be the set of n random lifetime from Weibull distribution with parameters λ and α.

The probability density function of Weibull distribution is:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data

where, the Cumulative Distribution Function (CDF) of the Weibull distribution is given as:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data

with λ as the scale parameter and α the shape parameter of the Weibull distribution.

Maximum likelihood estimation based on type-I censored data: The likelihood function as in the study of Klein and Moeschberger (2003) is:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(1)

  where, δi is 1 for failure and δi is 0 for censored observation and S(.) is the survival function.

 

The logarithm of the likelihood function can be expressed as follows:


Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(2)

To obtain the equations for the unknown parameters, we differentiate Eq. 2 partially with respect to the parameters λ and α and equal it to zero. The resulting equations are given as follows:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(3)

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(4)

Let Eq. 3 equals to zero, then the maximum likelihood estimator for the scale parameter of Weibull distribution is:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(5)

The shape parameter cannot be solved analytically and for that we use the Newton Raphson method to find the numerical solution following (Hossain and Zimmer, 2003).

The estimate of the survival function for Maximum Likelihood Estimation of Weibull distribution is:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(6)

The estimate of the hazard function of Weibull distribution given shape by Maximum Likelihood Estimation is as follows:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(7)

Maximum likelihood estimation based on type-II censored data: The likelihood function is:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(8)

The logarithm of the likelihood function can be expressed as follows:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(9)

To obtain the equations for the unknown parameters, Eq. 9 is differentiated partially with respect to the parameters λ and α and equal it to zero. The resulting equations are given, respectively as:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(10)

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(11)

Let Eq. 10 equals to zero, then the maximum likelihood estimator for the scale parameter of Weibull distribution is:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(12)

The shape parameter cannot be solved analytically and for that we employed Newton Raphson method.

The estimate of the survival function for MLE based type-II censored data:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(13)

The estimate of the hazard function of Weibull based type-II censored data:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(14)

Bayesian estimation of weibull based on type-I censored data: The Bayesian estimator with Jeffreys prior is described here, where Jeffreys prior is the square root of the determinant of the Fisher information matrix:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data

Then the Jeffreys prior is:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(15)

where, k is a constant.

The posterior probability density function of λ given the data (t1,…, tn) is obtained by dividing the joint probability density function with the marginal density function (Ahmed et al., 2012):

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(16)

With this, the Bayesian estimates for the scale parameter of Weibull distribution under squared error loss function are given as:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(17)

The estimator for survival function of Weibull distribution given shape by Bayesian using Jeffreys prior obtained as:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(18)

The estimator for hazard function of Weibull distribution given shape by Bayesian using Jeffreys prior obtained as:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(19)

Extension of Jeffreys prior information based on type-I censored data: Extension of Jeffreys prior information is the Fisher information with the variable c where, c is a positive real number as:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(20)

The posterior probability density function of λ based on type-I censored data is:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(21)

With this, the Bayesian for extension of Jeffreys prior estimates the scale parameter of Weibull distribution under squared error loss function are given as:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(22)

The estimator for survival function of Weibull distribution given shape by Bayesian using extension of Jeffreys obtained as:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(23)

The estimator for hazard function of Weibull distribution by Bayesian using extension of Jeffreys obtained as:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(24)

Bayesian estimation of Weibull based on type-II censored data: The posterior probability density function of λ based on type-II censored data:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(25)

With this, the Bayesian estimate for the scale parameter of Weibull distribution under squared error loss function is given as:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(26)

The estimate of the survival function of Weibull distribution given shape by Bayesian using Jeffreys prior is:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(27)

The estimate of the hazard function of Weibull distribution based on type-II censored data by Bayesian using Jeffreys prior is obtained by:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(28)

The posterior probability density function of λ based on type-II censored data:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(29)

With this, Bayesian with extension of Jeffreys prior estimate of the scale parameter of Weibull distribution based on type-II censored data are given as:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(30)

The estimate of the survival function of Weibull distribution given shape by Bayesian using extension of Jeffreys prior is obtained as:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(31)

The estimate of the hazard function of Weibull distribution by Bayesian using extension of Jeffreys prior based on type-II censored data obtained as:

Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data
(32)

Simulation study: To assess the performance of the maximum likelihood and Bayesian estimation based on type-I and II censored data to estimate the scale parameter follow by estimate survival and hazard functions. The Mean Squared Errors (MSE) was calculated using 10,000 replications for sample size n = 25, 50 and 100 of Weibull distribution with type-I and II censored data for different value of parameters were the scale parameter λ = 0.8 and 1.2, the shape parameter α = 0.5 and 1.5 and the two values of Jeffreys extension were c = 0.4 and 1.4, the considered values of parameters and extension of Jeffreys are meant for illustration only and other values can be taken for generating the samples from Weibull distribution.

RESULTS AND DISCUSSION

Four values of the estimators which are MLE, Bayesian using Jeffreys prior (BJ), extension of Jeffreys prior with c = 0.4 (BE(c = 0.4)) and extension of Jeffreys prior with c = 1.4 (BE(c = 1.4)) are shown in each column for each size.

Table 1 contains the estimate of the scale parameter of Weibull distribution based type-I censored data with maximum likelihood, Bayesian using Jeffreys prior and extension of Jeffreys prior information.

From Table 2 the results show that when c = 0.4, the maximum likelihood is better compared to the others with respect to the MSE of scale parameter. On the other hand, when c = 1.4, extension of Jeffrey is better compared to the others (Ahmed et al., 2011). This implies that, as the value of extension of Jeffreys prior is kept below one, it exerts very minimal influence on the posterior distribution but as it increases to at least above one, the influence becomes significant on the posterior distribution from which Bayesian inference is drawn to give a very small mean squared error as compared to maximum likelihood and that of Jeffreys prior.

Table 3 shows the comparison of the survival function with the maximum likelihood is noticed to be better compared to the others when c = 0.4. On the other hand, when c = 1.4, extension of Jeffrey is better compared to the others following Ahmed et al. (2011).

Table 1: Estimate λ based type-I censored data
Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data

Table 2: MSE for the scale parameter of Weibull based type-I censored data
Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data

Table 3: MSE for the survival function of Weibull based type- I censored data
Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data

Table 4: MSE for the hazard function of Weibull based type-I censored data
Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data

As shown in Table 4 the hazard function estimates based on type-I censoring data was compared. The results show that when c = 0.4, the maximum likelihood is better compared to the others. Nevertheless, when c = 1.4, extension of Jeffrey has the smaller mean squared error as compared to others.

Table 5 estimates the scale parameter of Weibull distribution based type-II censored data with Maximum likelihood, Bayesian using Jeffreys prior and extension of Jeffreys prior.

Table 6 observed that, the mean squared errors of the Bayes estimator based type-II censored data under the extension of Jeffreys prior (c = 0.4) is the best estimator of the scale parameter for all the cases.

The survival function as indicated in Table 7 show that, the Bayesian using extension of Jeffreys based on type-II censored data is the best for all cases.

Table 8 show that, extension of Jeffreys prior based on type-II censored data is the best estimation.

Table 5: Estimate λ based type-II censored data
Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data

Table 6: MSE for the scale parameter of Weibull based type-II censored data
Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data

Table 7: MSE for the survival function of Weibull based type-II censored data
Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data

Table 8: MSE for the hazard function of Weibull based type-II censored data
Image for - Bayesian Using Extension of Jeffreys Estimator of Weibull Distribution Based on Type-I and II Cnsored Data

CONCLUSION

In this study we have considered the Bayesian using extension of Jeffreys prior based on type-I and II censored data. The result show that, the Bayesian using extension of Jeffreys prior based on type-I is better estimator for all estimated of the scale parameter and the survival and hazard functions when the value of extension of Jeffreys is 1.4. On the other hand, the maximum likelihood method is better than others when the value of extension of Jeffreys is 0.4. The Bayesian using extension of Jeffreys prior based on type-II censored data is the best estimation compared to the other estimation.

REFERENCES

1:  Ahmed, A.O.M., N.A. Ibrahim, M.B. Adam and J. Arasan, 2012. Bayesian survival and hazard estimate for weibull censored time distribution. J. Applied Sci., 12: 1313-1317.
CrossRef  |  Direct Link  |  

2:  Hossain, A.M. and W.J. Zimmer, 2003. Comparison of estimation methods for Weibull parameters: Complete and censored samples. J. Stat. Comput. Simul., 73: 145-153.
CrossRef  |  Direct Link  |  

3:  Klein, J.P. and M.L. Moeschberger, 2003. Survival Analysis: Techniques for Censored and Truncated Data. 2nd Edn., Springer, New York, USA., ISBN-13: 9780387953991, pp: 75

4:  Kundu, D. and H. Howlader, 2010. Bayesian inference and prediction of the inverse Weibull distribution for Type-II censored data. Comput. Statist. Data Anal., 54: 1547-1558.
CrossRef  |  Direct Link  |  

5:  Kantar, Y.M. and B. Senoglu, 2008. A comparative study for the location and scale parameters of the Weibull distribution with given shape parameter. Comput. Geosci., 34: 1900-1909.
CrossRef  |  

6:  Pandey, B.N., N. Dwividi and B. Pulastya, 2011. Comparison between Bayesian and maximum likelihood estimation of the scale parameter in Weibull distribution with known shape under Linex loss function. J. Sci. Res., 55: 163-172.
Direct Link  |  

7:  Singh, U., P.K. Gupta and S.K. Upadhyay, 2005. Estimation of parameters for Exponentiated-Weibull family under type-II censoring scheme. Comput. Stat. Data Anal., 48: 509-523.
CrossRef  |  Direct Link  |  

8:  Sinha, S.K., 1986. Bayes estimation of the reliability function and hazard rate of a Weibull failure time distribution. Trabajos Estadistica, 1: 47-56.
CrossRef  |  Direct Link  |  

9:  Smith, R.L. and J.C. Naylor, 1987. A comparison of maximum likelihood and bayesian estimators for the three- parameter weibull distribution. Applied Stat., 36: 358-369.
CrossRef  |  Direct Link  |  

10:  Soliman, A.A., A.H. Abd Ellah and K.S. Sultan, 2006. Comparison of estimates using record statistics from weibull model: Bayesian and non-bayesian approaches. Comput. Stat. Data Anal., 51: 2065-2077.
CrossRef  |  Direct Link  |  

11:  Sun, D., 1997. A note on noninformative priors for Weibull distributions. J. Stat. Plann. Inference, 61: 319-338.
CrossRef  |  Direct Link  |  

12:  Ahmed, A.O.M., N.A. Ibrahim, J. Arasan and M.B. Adam, 2011. Extension of Jeffreys's prior estimate for weibull censored data using Lindley's approximation. Aust. J. Basic Applied Sci., 5: 884-889.
Direct Link  |  

©  2021 Science Alert. All Rights Reserved