Simple expressions for a bivariate chisquare distribution

The joint distribution of the estimated variances from a correlated bivariate normal distribution has a long history. However, its joint probability density function, conditional moments and product moments are only known as infinite series. In this paper, simpler expressions, mostly finite sums of elementary functions, are derived for these properties. Expressions are also derived for the joint moment generating function and the joint characteristic function.     

Admissible tests for exponential families with finite support 2

We state some necessary and sufficient conditions for admissibility of tests for a simple and a composite null hypotheses against ”one-sided” alternatives on multivariate exponential distributions with discrete support.

Admissibility of the maximum likelihood test for “one –sided” alternatives and z χ²test for the independence hypothesis in r× scontingency tables is deduced among others.     

On a distributionfree method in discriminant analysis

Linear discriminant rules for two symmetrical distributions, which only need the first and second moments of these distributions, are presented. The rules are based on Zhezhel's idea using the most unfavourable probabilities of misclassification as an optimality criterion. Also a rule is considered which deals with distributions differing in a location and scale parameter.     

Confidence intervals for a bounded mean in exponential families

Setting confidence bounds or intervals for a parameter in a restricted parameter space is an important issue in applications and is widely discussed in the recent literature. In this article, we focus on the distributions in the exponential families, and propose general forms of the truncated Pratt interval and rp interval for the means. We take the Poisson distribution as an example to illustrate the method and compare it with the other existing intervals. Besides possessing the merits from the theoretical inferences, the proposed intervals are also shown to be competitive approaches from simulation and real-data application studies.     

Sampling strategy for optimal classification into one of two correlated normal populations

A unit ω is to be classified into one of two correlated homoskedastic normal populations by linear discriminant function known as W classification statistic [T.W. Anderson, An asymptotic expansion of the distribution of studentized classification statistic, Ann. Statist. 1 (1973), pp. 964–972; T.W. Anderson, An Introduction to Multivariate Statistical Analysis, 2nd edn, Wiley, New York, 1984; G.J. Mclachlan, Discriminant Analysis and Statistical Pattern Recognition, John Wiley and Sons, New York, 1992]. The two populations studied here are two different states of the same population, like two different states of a disease where the population is the population of diseased patient. When a sample unit is observed in both the states (populations), the observations made on it (which form a pair) become correlated. A training sample is unbalanced when not all sample units are observed in both the states. Paired and also unbalanced samples are natural in studies related to correlated populations. S. Bandyopadhyay and S. Bandyopadhyay [Choosing better training sample for classifying an individual into one of two correlated normal populations, Calcutta Statist. Assoc. Bull. 54(215–216) (2003), pp. 167–180] studied the effect of unbalanced training sample structure on the performance of W statistics in the univariate correlated normal set-up for finding optimal sampling strategy for a better classification rate. In this study, the results are extended to the multivariate case with discussion on application in real scenario.     

On a sampling scheme with inclusion probability proportional to size

A sampling scheme for selection of a sample of two units with inclusion probability proportionalto size is suggested which provides a non–negative variance estimator of the variance of Horvitz–Thompson estimator. The suggested sampling scheme is shown to perform better than many of the existing unequal probability and inclusion probability proportional to size sampling Achemes for a number of natural populations.     

Exact Size and Power Properties of Five Tests for Multinomial Proportions

Estimators of the expectations of order statistics, suggested by Harrell &; Davis, are used in place of the order statistics in a quantile estimator, proposed by Kappenman , to produce a modification of Kappenman's procedure. Simulation studies indicate that the modification generally results in a reduction of mean squared error     

Comparison of Efficiency of Stratified and Unequal Probability Sampling

In this article, a choice of the optimum sampling design to study a finite population is studied. Three sampling schemes are compared, viz., Sunter's procedure of unequal probability sampling, stratified sampling under optimum stratification, and simple random sampling without replacement. The comparison is made against a background of various correlation between stratification and survey variables and various variability in the variables. Under weak correlation and large variability, stratification appeared to be more efficient than Sunter's procedure. Under strong correlation and/or low variability in the variables, the latter procedure was the most efficient. Simple random sampling was usually the least efficient.     

Robustness of R-Chart to Non Normality

This study addresses the appropriate d ₃ values for constructing range control charts (R-charts) when the distributions of the processes are the uniform, triangular, exponential, and Erlang. Comparisons of the range charts are based on Type I error probabilities obtained using simulations. The results reveal that inappropriate use of the d ₃ values strongly affected the performance of the R-charts. Practitioners should be more careful in selecting suitable coefficients when using R-charts methods to process data. The distribution of the processes must be examined before the coefficients are chosen.     

Importance Sampling for Sums of Lognormal Distributions with Applications to Operational Risk

In this paper we use Importance Sampling to estimate tail probabilities for a finite sum of lognormal distributions. We use a defensive mixture, and develop a method of choosing the parameters via the EM algorithm; we also consider the technique which assumes the importance sampling density to belong to the same parametric family of the random variables to be summed. In both cases, the instrumental density is found by minimizing Cross-Entropy. A comparison based on several simulation experiments shows that the defensive mixture has the best performance. Finally, we study the Poisson-lognormal compound distribution framework and present a real-data application.