Journal of Statistical Research of Iran
Volume:12 Issue: 2, 2016

In this paper, a skew normal state space model of RC electrical circuit is presented by considering the stochastic differential equation of the this circuit as the dynamic model with colored and white noise and considering a skew normal distribution instead of normal as the measurement noise distribution. Optimal filtering technique via sequential Monte Carlo perspective is developed for tracking the charge as the hidden state of this model. Furthermore, it is assumed that this model contains unknown parameters (resistance, capacitor, mean, variance and shape parameter of the skew normal as the measurement noise distribution). Bayesian framework is applied for estimation of both the hidden charge and the unknown parameters using particle marginal Metropolis-Hastings scheme. It is shown that the coverage percentage of skew normal is more than the one of normal as the measurement noise. Some simulation studies are carried out to demonstrate the efficiency of the proposed approaches.

In this paper, we use the method of modified signed log-likelihood ratio test for the problem of testing the equality of correlation coefficients in two independent bivariate normal distributions. We compare this method with two other approaches, Fisher's Z-transform and generalized test variable, using a Monte Carlo simulation. It indicates that the proposed method is better than the other approaches, in terms of the actual sizes and powers especially when the sample sizes are unequal. We illustrate performance of the proposed approach, using a real data set.

One of the most interesting problems in distribution theory is constructing the distributions, which are appropriate for fitting skewed and heavy-tailed data sets. In this paper, we introduce a skew-slash distribution by using the scale mixture of the generalized gamma distribution. Some properties of this distribution are obtained. An EM-type algorithm is presented to estimate the parameters. Finally, we provide a simulation study and an application to real data to illustrate the modeling strength of the proposed distribution.

Statistical prediction analysis plays an important role in a wide range of fields. Examples include engineering systems, design of experiments, etc. In this paper, based on progressively Type-II right censored data, Bayesian two-sample point and interval predictors are developed under both informative and non-informative priors. By assuming a generalized exponential model, prediction bounds as well as Bayes point predictors are obtained under the squared error loss (SEL) and the Linear-Exponential (LINEX) loss functions for the order statistic in a future progressively Type-II censored sample with an arbitrary progressive censoring scheme. The derived results may be used for prediction of total time on test in lifetime experiments. %in reliability analyses In addition to numerical method, Gibbs sampling procedure (as Markov Chain Monte Carlo method) are used to assess approximate prediction bounds and Bayes point predictors under the SEL and LINEX loss functions. The performance of the proposed prediction procedures are also demonstrated via a Monte Carlo simulation study and an illustrative example, for each method.

In this paper, a discrete analog of the beta-Rayleigh distribution is studied. This new distribution contains the generalized discrete Rayleigh and discrete Rayleigh distributions as special sub-models. Some distributional and moment properties of the new discrete distribution as well as its order statistics are discussed. We will see that the hazard rate function of the new model can be increasing, bathtub-shaped and upside-down bathtub. Estimation of the parameters is illustrated and, finally, the model with a real data set is examined.

In the present paper, we study shrinkage testimation for the unknown scale parameter $theta>0$ of the exponential distribution based on record data under the asymmetric squared log error loss function. A minimum risk unbiased estimator within the class of the estimators of the form $cT_m$ is derived, where $T_m$ is the maximum likelihood estimate of $theta$. Some shrinkage testimators are proposed and their risks are computed. The relative efficiencies of the shrinkage testimators with respect to a minimum risk unbiased estimator of the form $cT_m$ under the squared log error loss function are calculated for the comparison purposes. An illustrative example is also presented.