arXiv subjects

Debasis Kundu

Publications and source records attributed to Debasis Kundu.

36 records · Page 2Linked to original sources

arXiv2018

A New Decision Theoretic Sampling Plan for Exponential Distribution under Type-I Censoring

In this paper a new decision theoretic sampling plan (DSP) is proposed for Type-I censored exponential distribution. The proposed DSP is based on a new estimator of the expected lifetime of an exponential distribution which always exists, unlike the usual maximum likelihood estimator. The DSP is a modification of the Bayesian variable sampling plan of \cite{Lam:1994}. An optimum DSP is derived in the sense that it minimizes the Bayes risk. In terms of the Bayes risks, it performs better than Lam's sampling plan and its performance is as good as the Bayesian sampling plan of \cite{LLH:2002}, although implementation of the DSP is very simple. Analytically it is more tractable than the Bayesian sampling plan of \cite{LLH:2002}, and it can be easily generalized for any other loss functions also. A finite algorithm is provided to obtain the optimal plan and the corresponding minimum Bayes risk is calculated. Extensive numerical comparisons with the optimal Bayesian sampling plan proposed by \cite{LLH:2002} are made. The results have been extended for three degree polynomial loss function and for Type-I hybrid censoring scheme.

math.ST↗

arXiv2018

On a General Class of Discrete Bivariate Distributions

In this paper we develop a very general class of bivariate discrete distributions. The basic idea is very simple. The marginals are obtained by taking the random geometric sum of a baseline distribution function. The proposed class of distributions is a very flexible class of distributions in the sense the marginals can take variety of shapes. It can be multimodal as well as heavy tailed also. It can be both over dispersed as well as under dispersed. Moreover, the correlation can be of a wide range. We discuss different properties of the proposes class of bivariate distributions. The proposed distribution has some interesting physical interpretations also. Further, we consider two specific base line distributions namely; Poisson and negative binomial distributions for illustrative purposes. Both of them are infinitely divisible. The maximum likelihood estimators of the unknown parameters cannot be obtained in closed form. They can be obtained by solving three and five dimensional non-linear optimizations problems, respectively. To avoid that we propose to use the method of moment estimators and they can be obtained quite conveniently. The analyses of two real data sets have been performed to show the effectiveness of the proposed class of models. Finally, we discuss some open problems and conclude the paper.

stat.ME↗

arXiv2018

Chirp-like model and its parameter estimation

We propose a chirp-like signal model as an alternative to a chirp model and a generalisation of the sinusoidal model, which is a fundamental model in the statistical signal processing literature. It is observed that the proposed model can be arbitrarily close to the chirp model. The propounded model is similar to a chirp model in the sense that here also the frequency changes linearly with time. However, the parameter estimation of a chirp-like model is simpler compared to a chirp model. In this paper, we consider the least squares and the sequential least squares estimation procedures and study the asymptotic properties of these proposed estimators. These asymptotic results are corroborated through simulation studies and analysis of four speech signal data sets have been performed to see the effectiveness of the proposed model, and the results are quite encouraging.

stat.ME↗

arXiv2018

On Bivariate Discrete Weibull Distribution

Recently, Lee and Cha (2015, `On two generalized classes of discrete bivariate distributions', {\it American Statistician}, 221 - 230) proposed two general classes of discrete bivariate distributions. They have discussed some general properties and some specific cases of their proposed distributions. In this paper we have considered one model, namely bivariate discrete Weibull distribution, which has not been considered in the literature yet. The proposed bivariate discrete Weibull distribution is a discrete analogue of the Marshall-Olkin bivariate Weibull distribution. We study various properties of the proposed distribution and discuss its interesting physical interpretations. The proposed model has four parameters, and because of that it is a very flexible distribution. The maximum likelihood estimators of the parameters cannot be obtained in closed forms, and we have proposed a very efficient nested EM algorithm which works quite well for discrete data. We have also proposed augmented Gibbs sampling procedure to compute Bayes estimates of the unknown parameters based on a very flexible set of priors. Two data sets have been analyzed to show how the proposed model and the method work in practice. We will see that the performances are quite satisfactory. Finally, we conclude the paper.

stat.ME↗

arXiv2018

On approximate least squares estimators of parameters on one-dimensional chirp signal

Chirp signals are quite common in many natural and man-made systems like audio signals, sonar, radar etc. Estimation of the unknown parameters of a signal is a fundamental problem in statistical signal processing. Recently, Kundu and Nandi \cite{2008} studied the asymptotic properties of least squares estimators of the unknown parameters of a simple chirp signal model under the assumption of stationary noise. In this paper, we propose periodogram-type estimators called the approximate least squares estimators to estimate the unknown parameters and study the asymptotic properties of these estimators under the same error assumptions. It is observed that the approximate least squares estimators are strongly consistent and asymptotically equivalent to the least squares estimators. Similar to the periodogram estimators, these estimators can also be used as initial guesses to find the least squares estimators of the unknown parameters. We perform some numerical simulations to see the performance of the proposed estimators and compare them with the least squares estimators and the estimators proposed by Lahiri et al., \cite{2013}. We have analysed two real data sets for illustrative purposes.

stat.AP↗

arXiv2018

Hierarchical EM algorithm for estimating the parameters of Mixture of Bivariate Generalized Exponential distributions

This paper provides a mixture modeling framework using the bivariate generalized exponential distribution. We study different properties of this mixture distribution. Hierarchical EM algorithm is developed for finding the estimates of the parameters. The algorithm takes very large sample size to work as it contains many stages of approximation. Numerical Results are provided for more illustration.

stat.CO↗

arXiv2018

Univariate and Bivariate Geometric Discrete Generalized Exponential Distributions

Marshall and Olkin (1997, Biometrika, 84, 641 - 652) introduced a very powerful method to introduce an additional parameter to a class of continuous distribution functions and hence it brings more flexibility to the model. They have demonstrated their method for the exponential and Weibull classes. In the same paper they have briefly indicated regarding its bivariate extension. The main aim of this paper is to introduce the same method, for the first time, to the class of discrete generalized exponential distributions both for the univariate and bivariate cases. We investigate several properties of the proposed univariate and bivariate classes. The univariate class has three parameters, whereas the bivariate class has five parameters. It is observed that depending on the parameter values the univariate class can be both zero inflated as well as heavy tailed. We propose to use EM algorithm to estimate the unknown parameters. Small simulation experiments have been performed to see the effectiveness of the proposed EM algorithm, and a bivariate data set has been analyzed and it is observed that the proposed models and the EM algorithm work quite well in practice.

stat.ME↗

arXiv2018

Inference on Weibull Parameters Under a Balanced Two Sample Type-II Progressive Censoring Scheme

The progressive censoring scheme has received considerable amount of attention in the last fifteen years. During the last few years joint progressive censoring scheme has gained some popularity. Recently, the authors Mondal and Kundu ("A new two sample Type-II progressive censoring scheme", arXiv:1609.05805) introduced a balanced two sample Type-II progressive censoring scheme and provided the exact inference when the two populations are exponentially distributed. In this article we consider the case when the two populations follow Weibull distributions with the common shape parameter and different scale parameters. We obtain the maximum likelihood estimators of the unknown parameters. It is observed that the maximum likelihood estimators cannot be obtained in explicit forms, hence, we propose approximate maximum likelihood estimators, which can be obtained in explicit forms. We construct the asymptotic and bootstrap confidence intervals of the population parameters. Further we derive an exact joint confidence region of the unknown parameters. We propose an objective function based on the expected volume of this confidence set and using that we obtain the optimum progressive censoring scheme. Extensive simulations have been performed to see the performances of the proposed method, and one real data set has been analyzed for illustrative purposes.

stat.AP↗

arXiv2017

Bayesian analysis of three parameter singular Marshall-Olkin bivariate Pareto distribution

This paper provides bayesian analysis of singular Marshall-Olkin bivariate Pareto distribution. We consider three parameter singular Marshall-Olkin bivariate Pareto distribution. We consider two types of prior - reference prior and gamma prior. Bayes estimate of the parameters are calculated based on slice cum gibbs sampler and Lindley approximation. Credible interval is also provided for all methods and all prior distributions. A data analysis is kept for illustrative purpose.

stat.ME↗

arXiv2017

Estimation of P(X > Y ) for Weibull distribution based on hybrid censored samples

A Hybrid censoring scheme is mixture of Type-I and Type-II censoring schemes. Based on hybrid censored samples, this paper deals with the in- ference on R = P(X > Y ), when X and Y are two independent Weibull distributions with different scale parameters, but having the same shape pa- rameter. The maximum likelihood estimator (MLE), and the approximate MLE (AMLE) of R are obtained. The asymptotic distribution of the maxi- mum likelihood estimator of R is obtained. Based on the asymptotic distribu- tion, the confidence interval of R can be derived. Two bootstrap confidence intervals are also proposed. We consider the Bayesian estimate of R, and propose the corresponding credible interval for R. Monte Carlo simulations are performed to compare the different proposed methods. Analysis of a real data set has also been presented for illustrative purposes.

stat.OT↗

arXiv2017

Order Restricted Bayesian Analysis of a Simple Step Stress Model

In this article we consider a simple step stress set up under the cumulative exposure model assumption. At each stress level the lifetime distribution of the experimental units are assumed to follow the generalized exponential distribution. We provide the order restricted Bayesian inference of the model parameters by considering the fact that the expected lifetime of the experimental units are larger in lower stress level. Analysis and the related results are extended to different censoring schemes also. The Bayes estimates and the associated credible intervals of the unknown parameters are constructed using importance sampling technique. We perform extensive simulation experiments both for the complete and censored samples to see the performances of the proposed estimators. We analyze two simulated and one real data sets for illustrative purposes. An optimal value of the stress changing time is obtained by minimizing the total posterior coefficient of variations of the unknown parameters.

stat.AP↗

arXiv2017

Multivariate Geometric Skew-Normal Distribution

Azzalini (1985) introduced a skew-normal distribution of which normal distribution is a special case. Recently Kundu (2014) introduced a geometric skew-normal distribution and showed that it has certain advantages over Azzalini's skew-normal distribution. In this paper we discuss about the multivariate geometric skew-normal distribution. It can be used as an alternative to Azzalini's skew normal distribution. We discuss different properties of the proposed distribution. It is observed that the joint probability density function of the multivariate geometric skew normal distribution can take variety of shapes. Several characterization results have been established. Generation from a multivariate geometric skew normal distribution is quite simple, hence the simulation experiments can be performed quite easily. The maximum likelihood estimators of the unknown parameters can be obtained quite conveniently using expectation maximization (EM) algorithm. We perform some simulation experiments and it is observed that the performances of the proposed EM algorithm are quite satisfactory. Further, the analyses of two data sets have been performed, and it is observed that the proposed methods and the model work very well.

stat.ME↗

arXiv2017

Point and Interval Estimation of Weibull Parameters Based on Joint Progressively Censored Data

The analysis of progressively censored data has received considerable attention in the last few years. In this paper we consider the joint progressive censoring scheme for two populations. It is assumed that the lifetime distribution of the items from the two populations follow Weibull distribution with the same shape but different scale parameters. Based on the joint progressive censoring scheme first we consider the maximum likelihood estimators of the unknown parameters whenever they exist. We provide the Bayesian inferences of the unknown parameters under a fairly general priors on the shape and scale parameters. The Bayes estimators and the associated credible intervals cannot be obtained in closed form, and we propose to use the importance sampling technique to compute the same. Further, we consider the problem when it is known apriori that the expected lifetime of one population is smaller than the other. We provide the order restricted classical and Bayesian inferences of the unknown parameters. Monte Carlo simulations are performed to observe the performances of the different estimators and the associated confidence and credible intervals. One real data set has been analyzed for illustrative purpose.

stat.ME↗

arXiv2017

On Generalized Progressive Hybrid Censoring in presence of competing risks

The progressive Type-II hybrid censoring scheme introduced by Kundu and Joarder (\textit{Computational Statistics and Data Analysis}, 2509-2528, 2006), has received some attention in the last few years. One major drawback of this censoring scheme is that very few observations (even no observation at all) may be observed at the end of the experiment. To overcome this problem, Cho, Sun and Lee (\textit{Statistical Methodology}, 23, 18-34, 2015) recently introduced generalized progressive censoring which ensures to get a pre specified number of failures. In this paper we analyze generalized progressive censored data in presence of competing risks. For brevity we have considered only two competing causes of failures, and it is assumed that the lifetime of the competing causes follow one parameter exponential distributions with different scale parameters. We obtain the maximum likelihood estimators of the unknown parameters and also provide their exact distributions. Based on the exact distributions of the maximum likelihood estimators exact confidence intervals can be obtained. Asymptotic and bootstrap confidence intervals are also provided for comparison purposes. We further consider the Bayesian analysis of the unknown parameters under a very flexible Beta-Gamma prior. We provide the Bayes estimates and the associated credible intervals of the unknown parameters based on the above priors. We present extensive simulation results to see the effectiveness of the proposed method and finally one real data set is analyzed for illustrative purpose.

stat.AP↗

arXiv2017

Interval Estimation of the Unknown Exponential Parameter Based on Time Truncated Data

In this paper we consider the statistical inference of the unknown parameter of an exponential distribution based on the time truncated data. The time truncated data occurs quite often in the reliability analysis for type-I or hybrid censoring cases. All the results available today are based on the conditional argument that at least one failure occurs during the experiment. In this paper we provide some inferential results based on the unconditional argument. We extend the results for some two-parameter distributions also.

stat.AP↗

arXiv2017

Bivariate Discrete Generalized Exponential Distribution

In this paper we develop a bivariate discrete generalized exponential distribution, whose marginals are discrete generalized exponential distribution as proposed by Nekoukhou, Alamatsaz and Bidram ("Discrete generalized exponential distribution of a second type", Statistics, 47, 876 - 887, 2013). It is observed that the proposed bivariate distribution is a very flexible distribution and the bivariate geometric distribution can be obtained as a special case of this distribution. The proposed distribution can be seen as a natural discrete analogue of the bivariate generalized exponential distribution proposed by Kundu and Gupta ("Bivariate generalized exponential distribution", Journal of Multivariate Analysis, 100, 581 - 593, 2009). We study different properties of this distribution and explore its dependence structures. We propose a new EM algorithm to compute the maximum likelihood estimators of the unknown parameters which can be implemented very efficiently, and discuss some inferential issues also. The analysis of one data set has been performed to show the effectiveness of the proposed model. Finally we propose some open problems and conclude the paper.

stat.ME↗

arXiv2016

A Bayesian Approach to Estimation of Speaker Normalization Parameters

In this work, a Bayesian approach to speaker normalization is proposed to compensate for the degradation in performance of a speaker independent speech recognition system. The speaker normalization method proposed herein uses the technique of vocal tract length normalization (VTLN). The VTLN parameters are estimated using a novel Bayesian approach which utilizes the Gibbs sampler, a special type of Markov Chain Monte Carlo method. Additionally the hyperparameters are estimated using maximum likelihood approach. This model is used assuming that human vocal tract can be modeled as a tube of uniform cross section. It captures the variation in length of the vocal tract of different speakers more effectively, than the linear model used in literature. The work has also investigated different methods like minimization of Mean Square Error (MSE) and Mean Absolute Error (MAE) for the estimation of VTLN parameters. Both single pass and two pass approaches are then used to build a VTLN based speech recognizer. Experimental results on recognition of vowels and Hindi phrases from a medium vocabulary indicate that the Bayesian method improves the performance by a considerable margin.

cs.SD↗

arXiv2016

A New Two Sample Type-II Progressive Censoring Scheme

Progressive censoring scheme has received considerable attention in recent years. In this paper we introduce a new type-II progressive censoring scheme for two samples. It is observed that the proposed censoring scheme is analytically more tractable than the existing joint progressive type-II censoring scheme proposed by Rasouli and Balakrishnan \cite{RB:2010}. It has some other advantages also. We study the statistical inference of the unknown parameters based on the assumptions that the lifetime distribution of the experimental units for the two samples follow exponential distribution with different scale parameters. The maximum likelihood estimators of the unknown parameters are obtained and their exact distributions are derived. Based on the exact distributions of the maximum likelihood estimators exact confidence intervals are also constructed. For comparison purposes we have used bootstrap confidence intervals also. It is observed that the bootstrap confidence intervals work very well and they are very easy to implement in practice. Some simulation experiments are performed to compare the performances of the proposed method with the existing one, and the performances of the proposed method are quite satisfactory. One data analysis has been performed for illustrative purposes. Finally we propose some open problems.

stat.ME↗