فهرست مطالب
Journal of Statistical Modelling: Theory and Applications
Volume:4 Issue: 1, Winter and Spring 2023
- تاریخ انتشار: 1402/12/26
- تعداد عناوین: 12
-
-
Pages 1-10Moments play an essential role in the characterization of statistical distributions and criteria such as dispersion, skewness, and kurtosis. This article is a dissection of the central moments of two-point and binomial distributions. First, we consider the Bernoulli distribution of the population and generalize the results. With a simple method, we present the condition that when the sample size is large, the structure of the sample central moment consists of random variables independent of standard normal or chi-square or a combination of both. In the obtained results, the role of points that have a probability of 1/2 is very influential in the limit distribution.Keywords: Binomial Distribution, Central Moment, Convergence In Probability
-
Pages 11-24Suicide is a complex issue that affects many regions globally, and the factors that contribute to it can differ based on geographical and cultural contexts. In this study, we examine the relationship between socioeconomic factors and suicide mortality rates across 31 provinces in Iran, using data from 2020. We employ spatial econometric methods to analyze the data, allowing us to explore the statistical relationships between economic models and regional science. Our analysis reveals a significant clustering of suicide mortality in some western provinces, as shown by the distribution map of suicide mortality by province. We also find that the unemployment rate has a significant impact on suicide mortality. These findings provide valuable information for developing effective prevention strategies.Keywords: Geographic Analysis, Spatial Autocorrelation, Spatial Econometrics, Spillover Effect, Suicide Mortality
-
Pages 25-44In this paper, we propose a unit distribution called the logit Gudermannian distribution and present various statistical properties of the proposed model. Six parameter estimation methods are explored in the quest to estimate the parameters of the proposed distribution. We determine which estimation methods provide better parameter estimates through simulation studies. The study shows that the logit Gudermannian distribution provides a better fit for the datasets used than other unit distributions. Consequently, the logit Gudermannian distribution is used to develop a parametric regression model for studying the relationship between a unit response variable and other exogenous variables. The new regression model's performance is compared to that of other existing regression models and found to be competitive.Keywords: Gudermannian Distribution, One-Parameter Distribution, Regression, Unit Distribution, Univariate Transformation
-
Pages 45-58In the real world, we may come across with zero-inflated or zero-deflated count data that have a very short-run autocorrelation. Integer-valued moving average processes are suitable for modeling these data. In this paper, a non-negative integer-valued moving average process of the first order with zero-modified geometric innovations is introduced. This model is called zero-modified geometric INMA(1) process which contains geometric INMA(1) process as a particular case. Some statistical properties of the process are obtained. The parameters of the model are estimated by the Yule-Walker method. Then, using the simulation study, we evaluate the performance of this estimators. Finally, the model is applied to two examples of real time series of the monthly number of rubella cases and the annually number of earthquakes magnitude 8.0 to 9.9. Then, we exhibit the ability of the model for fitting and predicting count data with excess and deficit of zeros.Keywords: INMA(1) Process, Zero-Deflated, Zero-Inflated, Zero-Modified Geometric Distribution
-
Pages 59-73Nonlinear regression models have widespread applications across diverse scientific disciplines. Achieving precise fitting of the optimal nonlinear model is essential, taking into account the biases inherent in Bayesian optimal design. This study introduces a Bayesian optimal design utilizing the Dirichlet process as a prior. The Dirichlet process is a fundamental tool in exploring Nonparametric Bayesian inference, providing multiple well-suited representations. The research paper presents a novel one-parameter model, termed the ``unit-exponential distribution", specifically designed for the unit interval. Additionally, a representation is employed to approximate the D-optimality criterion, considering the Dirichlet process as a functional tool. Through this approach, the aim is to identify a nonparametric Bayesian optimal design.Keywords: Bayesian Optimal Design, D-Optimal Design, Dirichlet Process, Nonparametric Bayesian, Stick-Breaking Prior, Unit Exponential Model
-
Pages 75-93This paper describes the point and interval estimation of the unknown parameters of modified Weibull distribution under the adaptive Type-II progressive censored samples. First, we obtain the maximum likelihood estimation of parameters. Because maximum likelihood estimations should be solved in numerical methods and cannot be derived in a closed form, the approximate maximum likelihood estimations of the parameters are achieved. Also, asymptotic confidence intervals are obtained by earning the asymptotic distribution of the parameters. Moreover, two bootstrap confidence intervals are derived. Second, the Bayesian estimation of parameters is approximated using the Markov chain Monte Carlo algorithm and Lindley's method. Furthermore, the highest posterior density credible intervals of the parameters are derived. Finally, the different proposed estimations have been compared by the simulation studies and one data set is analyzed to illustrative aims.Keywords: Adaptive Type-II Progressive Censored Samples, Approximate Maximum Likelihood Estimation, Markov Chain Monte Carlo Algorithm, Modified Weibull Distribution
-
Pages 95-106This study focuses on estimating the reliability of a multicomponent stress-strength model using two Bayesian approaches: E-Bayesian and hierarchical Bayesian. This model follows Inverse Rayleigh distributions with distinct parameters. Additionally, the efficiency of the proposed methods is compared by employing Monte Carlo simulation and analyzing a data set.Keywords: E-Bayesian Estimation, Hierarchical Bayesian Estimation, Inverse Rayleigh Distribution, Multicomponent Stress-Strength Model, Reliability
-
Pages 107-127Often, reliability systems suffer shocks from external stress factors, stressing the system at random. These random shocks may have non-ignorable effects on the reliability of the system. In this paper, we provide sufficient and necessary conditions on components' lifetimes and their survival probabilities from random shocks for comparing the lifetimes of two fail-safe systems in two cases: (i) when components are independent, and then (ii) when components are dependent. We then apply the results for some distribution-free random variables with possibly different parameters to illustrate the established results.Keywords: Archimedean Copula, Distribution-Free Random Variables, Fail-Safe Systems, Random Shocks, Stochastic Orders
-
Pages 129-136For the interference models, the assumption is often made that the size of the blocks (k) is not greater than the number of treatments (t). Typically, it is difficult to specify optimal block designs theoretically or algorithmically when k> t. In this article, we focus on a one-sided interference model with compound symmetry correlation for the observations and obtain universally optimal block designs for both cases k≤ t and k>t. We present some methods for constructing these optimal designs for various numbers of treatments and block sizes.Keywords: Circular Block Design, Compound Symmetry Structure, Interference Model, Neighbor Balanced Design, Universal Optimality
-
Pages 137-146Spatial data analysis methods have many applications in various fields, such as agriculture, mining engineering, and meteorology. In this study, ordinary kriging and indicator kriging are considered to predict alumina grade in the Jajarm mine in Iran, and the precision of the methods is computed. A conditional simulation is carried out based on the data set for a more general comparison of ordinary and indicator kriging to interpolate Alumina grade in the mine. In the case of monitoring possible variation related to sample size and type of variogram model, simulations are performed with various sample sizes and different types of variogram models. Then ordinary and indicator kriging methods are applied for every set of simulated data (concerning different sample sizes and types of variogram models), and root of standardized mean square error prediction is considered as a cross-validation criterion to compare the kriging methods. The simulation results show that under the assumptions, ordinary kriging has better performance than the indicator kriging method.Keywords: Cross-Validation, Indicator Kriging, Ordinary Kriging, Random Field, Root Of Standardized Mean Squared Errors Prediction, Spatial Dat
-
Pages 147-156Most of the research on optimal designs concentrates on D-optimal designs for linear and nonlinear models with fixed effects. Recently nonlinear models with random effects have been of great attention because these models are more applicable to describing real data. In this paper, E-optimal designs for the Poisson regression with random effects have been considered. A new version of the equivalence theorem is prepared for this criterion in the Poisson regression model with random effects.Keywords: E-Optimal Design, Poisson Regression Model, Quasi-Likelihood Method, Random Effect
-
Pages 157-172In this paper, we develop a version of the weighted Marshall-Olkin bivariate exponential model by incorporating a new parameter. This parameter describes the dependence structure between margins via a copula function. We choose the inference for margins method to estimate the model parameters along with the copula parameter, as this method offers more advantages than the maximum likelihood estimation method. Additionally, we conduct a comprehensive simulation study to investigate the behavior of the copula parameter estimator and the remaining parameters. Finally, an analysis of a real dataset on automobile insurance reveals that the Clayton copula characterizes the dependence structure within the Archimedean copula familyKeywords: Archimedean Copula, Dependency, Inference For Margins Method, Optimization, Weighted Marshall-Olkin Bivariate Exponential