Reliability Estimation of Inverse Lomax Distribution Using Extreme Ranked Set Sampling
In survival analysis, the two-parameter inverse Lomax distribution is an important lifetime distribution. In this study, the estimation of is investigated when the stress and strength random variables are independent inverse Lomax distribution. Using the maximum likelihood approach, we obtain the estimator via simple random sample (SRS), ranked set sampling (RSS), and extreme ranked set sampling (ERSS) methods. Four different estimators are developed under the ERSS framework. Two estimators are obtained when both strength and stress populations have the same set size. The two other estimators are obtained when both strength and stress distributions have dissimilar set sizes. Through a simulation experiment, the suggested estimates are compared to the corresponding under SRS. Also, the reliability estimates via ERSS method are compared to those under RSS scheme. It is found that the reliability estimate based on RSS and ERSS schemes is more efficient than the equivalent using SRS based on the same number of measured units. The reliability estimates based on RSS scheme are more appropriate than the others in most situations. For small even set size, the reliability estimate via ERSS scheme is more efficient than those under RSS and SRS. However, in a few cases, reliability estimates via ERSS method are more accurate than using RSS and SRS schemes.
The inverse Lomax (ILo) distribution is considered as the reciprocal of the Lomax distribution. In some situations, it is a good alternative to the famous distributions like gamma, inverse Weibull, and Weibull. It has varied applications in modelling several types of data, including economics and actuarial sciences (see ). It has an application in geophysical databases . The ILo distribution has an important application in reliability analysis . Statistical inference for this distribution has been discussed by several researchers (see, for example, [4, 5]). In the present work, the ILo distribution is taken under the stress strength (S-S) model associated with any system that depends on different sampling schemes. The cumulative distribution function (cdf) of the ILo distribution with shape parameter and scale parameter is specified by the following:
The probability density function (pdf) of the ILo distribution is as follows:
The RSS was first introduced in  as a sampling scheme. The RSS scheme is used in situations when it is difficult and expensive to measure a large number of elements, but visually (without inspection) ranking some of them is easier and cheaper. This sampling design is both a cost-effective and powerful alternative to the commonly used SRS. This scheme involves randomly selecting sets (each of size elements) from the study population. The elements of each set are ordered with respect to the variable of the study by any negligible cost method or visually without measurements. Finally, the th minimum from the th set, , is specified for measurement. The obtained sample is called a RSS of set size . The whole procedure can be repeated times to yield a RSS of size . The mathematical theory of the RSS method has been provided in . Studies on RSS scheme have been proposed by several authors (see, for example, [8–15]).
Several modifications of the RSS have been proposed to improve the efficiency of the estimators. Herein, we are interested in the RSS and ERSS, presented in . The ERSS procedure involves randomly selecting sets (each of size m1 elements). The elements of each set are ordered with respect to variable of the study by visual inspection or any other cost free method. For an odd set size (OSZ), we select from the first samples the smallest ranked unit, from the other the largest ranked unit, and for the last sample select the median of the sample for actual measurement. For even set size (ESZ), we chose from samples the smallest ranked unit and from the other samples the largest ranked unit for actual measurement. This procedure can be repeated times to obtain units from ERSS data.
The S-S reliability is the probability of the system working when a strength is greater than a stress . So, the system will stop working when the applied stress is greater than its strength. Thus, the parameter is a measure of a system’s reliability, which has many applications in physics, engineering, genetics, psychology, and economics. There is an extensive literature on estimating based on SRS (see, for instance, [17–24]). However, in recent years, statistical inferences about the S-S model based on the RSS method have been considered by several researchers. Reference  discussed estimation of S-S reliability for exponential populations. Reference  proposed three estimators of when and are independent exponential populations. References [11, 27] discussed the estimation of the S-S model when and are two independent Burr type XII distribution under several modifications of the RSS method. Estimation of the S-S model for Weibull and Lindley distributions has been discussed, respectively, in [28, 29]. Reference  obtained a reliability estimator of for the exponentiated Pareto distribution under the RSS scheme.
The S-S model is one of the important approaches in reliability analysis. The S-S model can be used to solve a variety of engineering problems, such as determining whether a building’s strength should be subjected to the design earthquake, whether a rocket motor’s strength should be greater than the operating pressure, and comparing the strength of different materials. The ILo is one of the distributions which is used quite effectively for modelling the strength of data used in economics, geography, actuarial, and medical fields. It has been discovered to be very flexible in analyzing situations with a realized nonmonotonic failure rate, which has wide applications in modelling life components. The RSS method and its modifications are frequently employed to gather samples that are more representative of the underlying population, when sampling units are expensive and difficult to measure but easy and inexpensive to arrange according to the variable of interest. In this method, ranking can be done using expert opinions, auxiliary variables, or any other low-cost approach. Statistical inference on the S-S model, based on the RSS scheme and its variations, has recently gotten a lot of attention. Due to the importance of the ILo distribution in reliability research, we propose to evaluate the reliability estimator of the S-S model where the strength and stress are both independent. Under SRS, RSS, and ERSS methods, the maximum likelihood (ML) estimators of are derived. Based on the ERSS scheme, we get the ML estimator of when both and populations have similar or dissimilar set sizes. We evaluate the accuracy of estimators using absolute biases (ABs), mean squared errors (MSEs), and relative efficiencies (REs) in a simulated exercise. The remainder of this essay is structured in the following manner. In Section 2, we extract ’s expression and use SRS to calculate ’s ML estimator. In Section 3, the RSS is used to obtain an estimator for the S-S model. Section 4 presents reliability estimators of the S-S model using ERSS methodology. A numerical analysis is included in Section 5. Finally, in Section 6, we bring the paper to a close.
2. Estimator of Using SRS
In this section, we derive the expression of as well as obtain its ML estimator. Assuming that the strength and stress are independently distributed random variables with the same scale parameter, where and , the system’s reliability with stress variable and strength variable is given by the following:
The strength-stress parameter given in (3) depends on the shape parameters and . Let be a SRS of size from the , and of size be SRS from the being independent with a common scale parameter. The log-likelihood of the observed sample is given by
The partial derivatives of with respect to , , and are, respectively, given by
3. Estimator of Using RSS
We derive the reliability estimator when the random samples of strength and stress are observed from the RSS design. Let be a RSS of size for where is the th order statistics of size of the th cycle.
Similarly, let be a RSS method of size , where is the set size and is the number of cycles. For simplified forms, we use the notations and instead of the notations and , respectively, for easy understanding and the simplicity. The pdf of and are given, respectively, by
The likelihood function, say , based on RSS is given by
The ML estimators of , , and are the solutions of the following equations:
As can be seen, we use iterative approaches to solve Equations (11)–(13) because there are no explicit solutions. As a result, the ML estimator of S-S reliability is obtained based on the invariance property of ML estimators.
4. Estimator of Using ERSS
In this section, we obtain the ML estimator of when strength and stress have an ILo distribution under the ERSS design. In these respects, the reliability estimator is considered in two cases when both and distributions have similar or dissimilar set sizes. We derive the reliability estimator when the random samples of strength and stress are observed from ERSS.
4.1. Estimator of
Herein, we derive the reliability estimator when the observed data of strength and stress populations are drawn from the ERSS scheme with OSZ. Suppose that where and are the ERSS scheme drawn from with sample size , where is the set size and is the number of cycles. Let , , and are the smallest, median, and largest order statistics from the th set of size of the th cycle, respectively. The observed ERSS with OSZ (for one cycle) is presented in Table 1.
The pdfs of the smallest, median, and largest order statistics from the th set of size of the th cycle are defined, respectively, as follows.
Similarly, assume that , where and are the ERSS drawn from with a sample size , where is the set size and is the number of cycles. Let , , and are the smallest, median, and largest order statistics from the th set of size of the th cycle, respectively. The pdfs of the smallest, median, and largest order statistics from the th set of size of the th cycle are defined, respectively, as follows:
The likelihood function, say , based on ERSS method with OSZ is given by the following.
The ML estimators of the parameters , , and are the solutions of the following equations: where . We obtain the parameter’s estimator by solving numerically Equations (21)–(23) using an iterative technique. As a result, the S-S reliability estimator is produced from (3).
4.2. Estimator of
Herein, we derive the reliability estimator when the observed data of strength and stress distributions are drawn from the ERSS method with ESZ. Let where and are the ERSS with ESZ drawn from with sample size . Let and are the smallest and largest order statistics from the th set of size of the th cycle, respectively. The observed ERSS with ESZ (for one cycle) is represented in Table 2.
The pdfs of and from the th set of size of the th cycle are defined in (14) and (16). Similarly, let , where and are the ERSS with ESZ drawn from with sample size . The pdfs of and from the th set of size of the th cycle are defined in (17) and (19). The likelihood function, say , based on ERSS with ESZ, is given by the following:
The ML estimators of , , and are the solutions of the following likelihood equations:
4.3. Estimator of
Here, we obtain the S-S reliability estimator when the observed samples of strength are drawn from ERSS with OSZ, while observed samples of stress are drawn from ERSS with ESZ. Let where and are the ERSS drawn from with sample size . The pdfs of , , and are provided in Equations (14)–(16).
Suppose that , where and are the ERSS with ESZ drawn from with sample size , where the density function of and are obtained in Equations (17) and (19). Hence, the likelihood function, say , in this case, is given by the following:
4.4. Estimator of
Here, we obtain the S-S reliability estimator when the observed samples of strength are drawn from ERSS with ESZ, while observed samples of stress are drawn from ERSS with OSZ.
Suppose that , where and are the ERSS with ESZ drawn from with sample size , with pdfs (14) and (16). Let , where and are the ERSS drawn from with sample size , with pdfs (17)–(19). Hence, the likelihood function, say , in this case, is given by the following:
5. Numerical Representation
This section introduces some simulations to assess how well the ML estimation of the S-S reliability function worked based on the proposed sampling scheme. A comparison is made between different estimates based on SRS, RSS, and ERSS methods. The following is a full description of the simulated experiment. (i)Using inverse transformation, 1000 random samples are created from the strength and stress distributions(ii)The parameter’s values are chosen as , , and the true value for the system reliability is determined as 0.909, 0,833, 0.714, and 0.625, respectively(iii)The sample sizes are selected as for SRS(iv)The number of cycles is set to be , while the set sizes are selected as . As a result, the sample sizes for RSS and ERSS sampling designs are determined as and (v)A numerical technique is utilized to obtain the ML of parameters and consequently the reliability estimate using the three sampling strategies(vi)The performance of the S-S reliability estimates for the three sampling strategies is evaluated using ABs, MSEs, and REs measures(vii)The AB is defined as: , where .(viii)Three REs of reliability estimates are provided and defined as follows: (ix)Tables 3–6 describe the reliability estimates , ABs, and MSEs based on SRS, RSS, and ERSS schemes. The REs of based on ERSS and RSS with respect to SRS and RSS for various sample sizes are presented in Tables 3–6
Tables 3–6 and Figures 1–6 show the following numerical results: (i)The reliability estimates via RSS are more efficient than the corresponding based on SRS based in the same number of measured units(ii)With the exception of , obtained by ERSS are more efficient than those obtained through SRS(iii)Except for , the MSEs of based on RSS technique are smaller than the corresponding via ERSS scheme at , as shown in Table 3(iv)Except for , the MSEs of based on RSS scheme are more efficient than the corresponding via ERSS at true value (Table 4)(v)At true value , the MSEs of based on RSS scheme are more efficient than the corresponding via ERSS except for as seen in Table 5(vi)At actual value , the MSEs of based on RSS scheme are more efficient than the corresponding ERSS, except for (see Table 6)(vii)The AB of RSS is smaller than SRS and ERSS in most of the cases(viii)Expect at true value for , the MSE of gets the largest value via SRS design and smallest values via RSS scheme (see Figure 1)(ix)Expect at true value , the MSE of based on ERSS scheme is smaller than those under RSS and SRS methods at (Figure 2)(x)The MSE of based on ERSS obtains the fewest values compared to the others under RSS and SRS at true value and (see Figure 3)(xi)Figure 4 indicates that, at , the MSE of based on the ERSS scheme is smaller than the comparable via RSS and SRS(xii)Figures 3 and 4 indicate that the MSE of decreases as the true value of increases(xiii)For the RE2 of based on ERSS scheme is more efficient than those under RSS and SRS except at true value , as shown in Figure 5(xiv)Figure 6 illustrates that at , the RE2 of the ERSS scheme is more efficient than those via RSS and SRS schemes with the exception of true value (xv)The MSEs of the S-S reliability estimate in all schemes decrease as the actual value of increases in most of the cases (see Figures 1–6).
This article tackles the estimation of the S-S reliability when the strength and stress are independent inverse Lomax distributed random variables. Maximum likelihood estimators of are computed using the SRS, RSS, and ERSS schemes. The reliability estimator is computed in four situations using ERSS design. Simulation research is conducted to evaluate the performance of the proposed estimates. From the simulation outcomes, it is observed that the MSEs of reliability estimates based on SRS data are bigger than the comparable based on RSS and ERSS data, respectively. In most cases, at , the MSEs of reliability estimates under ERSS are the shortest when compared to similar estimators based on RSS and SRS data. The efficiency of all estimates improves as the actual value of reliability increases in almost all cases. This study showed that the reliability estimates based on RSS are more efficient than those based on ERSS and SRS. For most actual values of , the reliability estimate via the ERSS technique is more efficient than those under RSS and SRS for small even set sizes. In some cases, estimates of reliability obtained by ERSS are more efficient than those obtained through RSS and SRS designs. In a future work, one may consider the problem of estimating based on double extreme ranked set sampling , modified robust extreme ranked set sampling , stratified quartile ranked set sampling , and multistage percentile and quartile ranked set samples methods [34, 35].
There is no data is included in the paper.
Conflicts of Interest
The authors declare that they have no conflicts of interest to report regarding the present study.
This project was supported by the King Saud University, Deanship of Scientific Research, College of Science Research Center.
C. Kleiber and S. Kotz, Statistical Size Distributions in Economics and Actuarial Sciences, vol. 470, John Wiley & Sons, Inc., Hoboken, New Jersey, 2003.View at: Publisher Site
D. McKenzie, C. Miller, and D. A. Falk, The Landscape Ecology of Fire, Springer Science & Business Media, NewYork, 2011.
A. S. Yadav, S. K. Singh, and U. Singh, “On hybrid censored inverse Lomax distribution: application to the survival data,” Statistica, vol. 76, no. 2, pp. 185–203, 2016.View at: Google Scholar
S. Singh, U. Singh, and A. Yadav, “Reliability estimation for inverse Lomax distribution under type Π censored data using Markov chain Monte Carlo method,” International Journal of Mathematics and Statistics, vol. 17, no. 1, pp. 128–146, 2016.View at: Google Scholar
M. F. Al-Saleh and K. Al-Shrafat, “Estimation of average milk yield using ranked set sampling,” Environmetrics: The Official Journal of the International Environmetrics Society, vol. 12, no. 4, pp. 395–399, 2001.View at: Google Scholar
A. S. Hassan, “Maximum likelihood and Bayes estimators of the unknown parameters for exponentiated exponential distribution using ranked set sampling,” International Journal of Engineering Research and Applications, vol. 3, no. 1, pp. 720–725, 2013.View at: Google Scholar
A. S. Hassan, S. M. Assar, and M. Yahya, “Estimation of R= P[Y< X] for Burr type XII distribution based on ranked set sampling,” International Journal of Basic and Applied Sciences, vol. 3, no. 3, pp. 274–280, 2014.View at: Google Scholar
A. I. Al-Omari and A. Haq, “Novel entropy estimators of a continuous random variable,” International Journal of Modeling, Simulation, and Scientific Computing, vol. 10, no. 2, article 1950004, 2019.View at: Google Scholar
F. Downton, “The estimation of Pr(Y<X) in the normal case,” Technometrics, vol. 15, no. 3, pp. 551–558, 1953.View at: Google Scholar
S. Kotz and M. Pensky, The Stress-Strength Model and Its Generalizations: Theory and Applications, World Scientific, Singapore, 2003.View at: Publisher Site
A. S. Hassan and D. Al-Sulami, “Estimation of P[Y<X] in the case of exponentiated Weibull distribution,” The Egyptian Statistical Journal, Faculty of Graduate Studies for Statistical Research, Cairo University, vol. 52, no. 2, pp. 76–95, 2008.View at: Google Scholar
S. Sengupta and S. Mukhuti, “Unbiased estimation of P(X>Y) for exponential populations using order statistics with application in ranked set sampling,” Communications in Statistics-Theory and Methods, vol. 37, no. 6, pp. 898–916, 2008.View at: Google Scholar
H. A. Muttlak, W. A. Abu-Dayyeh, M. F. Saleh, and E. Al-Sawi, “Estimating P(Y<X) using ranked set sampling in case of the exponential distribution,” Communications in Statistics-Theory and Methods, vol. 39, no. 10, pp. 1855–1868, 2010.View at: Google Scholar
A. S. Hassan, S. M. Assar, and M. Yahya, “Estimation of P(Y<X) for Burr distribution under several modifications for ranked set sampling,” Australian Journal of Basic and Applied Sciences, vol. 9, no. 1, pp. 124–140, 2015.View at: Google Scholar
A. I. Al-Omari, I. M. Almanjahie, A. S. Hassan, and H. F. Nagy, “Estimation of the stress–strength reliability for exponentiated Pareto distribution using median and ranked set sampling methods,” CMC-Computers, Materials and Continua, vol. 64, no. 2, pp. 835–857, 2020.View at: Publisher Site | Google Scholar
A. I. Al-Omari and K. Jaber, “Improvement in estimating the population mean in double extreme ranked set sampling,” International Mathematical Forum, vol. 5, no. 26, pp. 1265–1275, 2010.View at: Google Scholar
A. I. Al-Omari, “Estimation of mean based on modified robust extreme ranked set sampling,” Journal of Statistical Computation and Simulation, vol. 81, no. 8, pp. 1055–1066, 2010.View at: Google Scholar
M. Syam, K. Ibrahim, and A. I. Al-Omari, “The efficiency of stratified quartile ranked set sample in estimating the population mean,” Tamsui Oxford Journal of Information and Mathematical Sciences, vol. 28, no. 2, pp. 175–190, 2012.View at: Google Scholar
A. A. Jemain and A. I. Al-Omari, “Multistage percentile ranked set samples,” Advances and Applications in Statistics, vol. 7, no. 1, pp. 127–139, 2007.View at: Google Scholar
A. A. Jemain and A. I. Al-Omari, “Multistage quartile ranked set samples,” Pakistan Journal of Statistics, vol. 23, no. 1, pp. 11–22, 2007.View at: Google Scholar