DOCSLIB.ORG

Stat 609: Mathematical Statistics Lecture 16

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Stat 609: Mathematical Statistics Lecture 16 Lecture 16: Hierarchical models and miscellanea It is often easier to model a practical situation by thinking of things in a hierarchy. Example 4.4.1 (binomial-Poisson hierarchy) An insect lays many eggs, each surviving with probability p. On the average, how many eggs will survive? Let Y be the number of eggs and X be the number of survivors; both are random variables. We can model this situation by first modeling the distribution of Y ; given Y , we then model the distribution of XjY . We can then obtain the joint distribution of (X;Y ) and marginal distributions of X and Y . We can model the number of eggs by a Poisson distribution, i.e., Y ∼ Poisson(l), where l > 0 is the average of eggs. Given Y , we can model the number of survivors as beamer-tu-logo XjY ∼ binomial(p;Y ). UW-Madison (Statistics) Stat 609 Lecture 16 2015 1 / 15 What is the marginal distribution of X? For x = 0;1;2;:::, ¥ P(X = x) = ∑ P(X = x;Y = y) x ≤ y y=x ¥ = ∑ P(X = xjY = y)P(Y = y) y=x ¥ y e−l l y = ∑ px (1 − p)y−x y=x x y! (lp)x e−l ¥ [(1 − p)l]y−x = ∑ x! y=x (y − x)! (lp)x e−l ¥ [(1 − p)l]t = ∑ t = y − x x! t=0 t! (lp)x e−l e−pl (lp)x = e(1−p)l = x! x! The distribution of X is Poisson(pl)! beamer-tu-logo UW-Madison (Statistics) Stat 609 Lecture 16 2015 2 / 15 Then, E(X) = pl, which can also be obtained using E(X) = E[E(XjY )] = E(pY ) = pE(Y ) = pl without using the marginal distribution of X. If we begin with our model by saying that X ∼ Poisson(q), then q = pl with Y playing no role at all. Introducing Y in the hierarchy was mainly aid our understanding of the model. Example (binomial-binomial hierarchy) A very similar hierarchical model can be described as follows. A market survey is conducted to study whether a new product is preferred over the product currently available in the market (old product). The survey is conducted by mail. Questionnaires are sent along with the sample products (both new and old) to N customers randomly selected from a population. Each customer is asked to fill out the questionnaire and returnbeamer-tu-logo it, with response 1 (new is better than old) or 0 (otherwise). UW-Madison (Statistics) Stat 609 Lecture 16 2015 3 / 15 Let X be the number of ones in the returned questionnaires. What is the distribution of X? If every customer returns the questionnaire, then (from elementary probability) X ∼ binomial(p;N) (assuming that the population is large enough so that customers respond independently), where p 2 (0;1) is the overall rate of customers who prefer the new product. Some customers, however, do not return the questionnaires. Let Y be the number of customers who respond. If customers respond independently with the same probability p 2 (0;1), then Y ∼ binomial(p;N). Given Y = y (an integer between 0 and N), XjY = y ∼ (p;y) if y ≥ 1 and the point mass at 0 if y = 0. For x = 0;1;:::;N, N N P(X = x) = P(X = x;Y = k) = P(X = xjY = k)P(Y = k) ∑ ∑ beamer-tu-logo k=x k=x UW-Madison (Statistics) Stat 609 Lecture 16 2015 4 / 15 N k N = ∑ px (1 − p)k−x pk (1 − p)N−k k=x x k N N N − xp − pp k−x 1 − p N−k = (pp)x (1 − pp)N−x ∑ x k=x k − x 1 − pp 1 − pp N = (pp)x (1 − pp)N−x x It turns out that the marginal distribution of X is the binomial(pp;N) distribution. Hierarchical models can have more than two stages. The advantage is that complicated processes may be modeled by a sequence of relatively simple models placed in a hierarchy. Conditional distributions play a central role. The random variables in hierarchical models may be all discrete (as in our previous examples), all continuous, or some discrete beamer-tu-logo and some continuous. UW-Madison (Statistics) Stat 609 Lecture 16 2015 5 / 15 More general joint pdf’s and conditional distributions So far we have considered only the situation where all random variables are continuous or all are discrete. What if X is a continuous random variable and Y is a discrete random variable on Y ? We can define the joint “pdf” to be a function f (x;y) satisfying Z x P(X ≤ x;Y = y) = f (t;y)dt; x 2 R; y 2 Y −¥ Then the marginal pdf of X and pmf of Y are respectively Z ¥ fX (x) = ∑ f (x;y) and fY (y) = f (x;y)dx y2Y −¥ The conditional pdf of XjY or pmf of Y jX can be defined as before. Similarly we can deal with the situation where X is a continuous random vector and Y is a discrete random vector. If X or Y is neither discrete nor continuous, we can still define conditional distributions. But it involves higher level mathematics. beamer-tu-logo UW-Madison (Statistics) Stat 609 Lecture 16 2015 6 / 15 Example. Suppose that X is a continuous random variable and Y is a discrete random variable with Y = f0;1g, and the joint pdf is 8 2 < pa e−x =2 y = 0; f (x;y) = 2p x 2 R 1−a −jxj : 2 e y = 1; where 0 < a < 1. Then a −x2=2 1 − a −jxj fX (x) = f (x;0) + f (x;1) = p e + e x 2 R 2p 2 ( Z ¥ a y = 0 fY (y) = f (x;y)dx = −¥ 1 − a y = 1 8 2 2 > pa e−x =2 pa e−x =2 + 1−a e−jxj y = 0 f (x;y) <> 2p 2p 2 f (yjx) = = fX (x) 2 > 1−a e−jxj pa e−x =2 + 1−a e−jxj y = 1 :> 2 2p 2 beamer-tu-logo UW-Madison (Statistics) Stat 609 Lecture 16 2015 7 / 15 8 1 −x2=2 f (x;y) < p e y = 0 f (xjy) = = 2p x 2 R fY (y) 1 −jxj : 2 e y = 1 Mixture distributions In the previous example, the pdf fX is referred to as a mixture distribution (or pdf), since it is a convex combination of two distributions (pdf’s). Example 4.4.5 (three-stage hierarchy) Consider a generalization of Example 4.4.1, where instead of one mother insect there are a large number of mothers and one mother is chosen at random from a (possibly continuous) distribution G. The following three-stage hierarchy may be more appropriate: number of survivors X from Y eggs XjY ∼ binomial(p;Y ) number of eggs Y from a given mother ZY jZ ∼ Poisson(Z ) a particular mother ZZ ∼ G beamer-tu-logo UW-Madison (Statistics) Stat 609 Lecture 16 2015 8 / 15 Suppose that Z ∼ exponential(b). Then E(X) = E[E(XjY )] = pE(Y ) = pE[E(Y jZ)] = pE(Z) = pb The three-stage model can be thought of as a two-stage model by combining the last two stages; we just need to obtain the marginal distribution of Y . Z ¥ Z ¥ −z y e z 1 −z=b P(Y = y) = f (yjz)fZ (z)dz = e dz 0 0 y! b ¥ 1 Z −1 1 1 = y −z(1+b ) = Γ( + ) z e dz y 1 −1 y+1 by! 0 by! (1 + b ) 1 b y+1 1 1 y = = 1 − b (1 + b)y+1 1 + b 1 + b which is the geometric distribution with probability (1 + b)−1, i.e., number of survivors X from Y eggs XjY ∼ binomial(p;Y ) number of eggs YY ∼ geometric((1 + b)−1) beamer-tu-logo However, the three-stage model is easier to understand. UW-Madison (Statistics) Stat 609 Lecture 16 2015 9 / 15 Calculation of mean and variance Aside from the advantage in understanding models, hierarchical models can often make the calculation of mean and variance easier. Let X be a random variable having a pdf that is the noncentral chi- square pdf with noncentrality parameter l and degrees of freedom n: ¥ xn=2+k−1e−x=2 (l=2)k e−l=2 f (x) = x > 0 X ∑ n=2+k k=1 Γ(n=2 + k)2 k! Calculating E(X) is directly using this pdf not easy. However, the pdf is the marginal pdf of X in the following two-stage model: XjK ∼ chi square with degrees of freedom n + 2K K ∼ Poisson(l=2) Then E(X) = E[E(XjK )] = E(n + 2K ) = n + 2E(K ) = n + l Var(X) = E[Var(XjK )] + Var(E(XjK )) = E(2n + 4K ) + Var(n + 2K ) beamer-tu-logo = 2n + 4E(K ) + 4Var(K ) = 2n + 4l UW-Madison (Statistics) Stat 609 Lecture 16 2015 10 / 15 Example 4.4.6. Suppose that XjP ∼ binomial(n;P); P ∼ beta(a;b) How to calculate E(X) and Var(X)? na E(X) = E[E(XjP)] = E(nP) = nE(P) = a + b Using the variance formula derived previously, Var(X) = Var(E(XjP)) + E[Var(XjP)] = Var(nP) + E[nP(1 − P)] = n2Var(P) + nE(P) − nE(P2) n2ab na na(a + 1) = + − (a + b)2(a + b + 1) a + b (a + b)(a + b + 1) nab(a + b + n) = (a + b)2(a + b + 1) beamer-tu-logo UW-Madison (Statistics) Stat 609 Lecture 16 2015 11 / 15 Theorem 4.7.9 (Covariance inequality) Let X be a random variable and g and h be nondecreasing functions such that E[g(X)], E[h(X)], and E[g(X)h(X)] exist.
Load more

Recommended publications
  • Chapter 6 Continuous Random Variables and Probability
    EF 507 QUANTITATIVE METHODS FOR ECONOMICS AND FINANCE FALL 2019 Chapter 6 Continuous Random Variables and Probability Distributions Chap 6-1 Probability Distributions Probability Distributions Ch. 5 Discrete Continuous Ch. 6 Probability Probability Distributions Distributions Binomial Uniform Hypergeometric Normal Poisson Exponential Chap 6-2/62 Continuous Probability Distributions § A continuous random variable is a variable that can assume any value in an interval § thickness of an item § time required to complete a task § temperature of a solution § height in inches § These can potentially take on any value, depending only on the ability to measure accurately. Chap 6-3/62 Cumulative Distribution Function § The cumulative distribution function, F(x), for a continuous random variable X expresses the probability that X does not exceed the value of x F(x) = P(X £ x) § Let a and b be two possible values of X, with a < b. The probability that X lies between a and b is P(a < X < b) = F(b) -F(a) Chap 6-4/62 Probability Density Function The probability density function, f(x), of random variable X has the following properties: 1. f(x) > 0 for all values of x 2. The area under the probability density function f(x) over all values of the random variable X is equal to 1.0 3. The probability that X lies between two values is the area under the density function graph between the two values 4. The cumulative density function F(x0) is the area under the probability density function f(x) from the minimum x value up to x0 x0 f(x ) = f(x)dx 0 ò xm where
    [Show full text]
  • Extremal Dependence Concepts
    Extremal dependence concepts Giovanni Puccetti1 and Ruodu Wang2 1Department of Economics, Management and Quantitative Methods, University of Milano, 20122 Milano, Italy 2Department of Statistics and Actuarial Science, University of Waterloo, Waterloo, ON N2L3G1, Canada Journal version published in Statistical Science, 2015, Vol. 30, No. 4, 485{517 Minor corrections made in May and June 2020 Abstract The probabilistic characterization of the relationship between two or more random variables calls for a notion of dependence. Dependence modeling leads to mathematical and statistical challenges and recent developments in extremal dependence concepts have drawn a lot of attention to probability and its applications in several disciplines. The aim of this paper is to review various concepts of extremal positive and negative dependence, including several recently established results, reconstruct their history, link them to probabilistic optimization problems, and provide a list of open questions in this area. While the concept of extremal positive dependence is agreed upon for random vectors of arbitrary dimensions, various notions of extremal negative dependence arise when more than two random variables are involved. We review existing popular concepts of extremal negative dependence given in literature and introduce a novel notion, which in a general sense includes the existing ones as particular cases. Even if much of the literature on dependence is focused on positive dependence, we show that negative dependence plays an equally important role in the solution of many optimization problems. While the most popular tool used nowadays to model dependence is that of a copula function, in this paper we use the equivalent concept of a set of rearrangements.
    [Show full text]
  • Joint Probability Distributions
    ST 380 Probability and Statistics for the Physical Sciences Joint Probability Distributions In many experiments, two or more random variables have values that are determined by the outcome of the experiment. For example, the binomial experiment is a sequence of trials, each of which results in success or failure. If ( 1 if the i th trial is a success Xi = 0 otherwise; then X1; X2;:::; Xn are all random variables defined on the whole experiment. 1 / 15 Joint Probability Distributions Introduction ST 380 Probability and Statistics for the Physical Sciences To calculate probabilities involving two random variables X and Y such as P(X > 0 and Y ≤ 0); we need the joint distribution of X and Y . The way we represent the joint distribution depends on whether the random variables are discrete or continuous. 2 / 15 Joint Probability Distributions Introduction ST 380 Probability and Statistics for the Physical Sciences Two Discrete Random Variables If X and Y are discrete, with ranges RX and RY , respectively, the joint probability mass function is p(x; y) = P(X = x and Y = y); x 2 RX ; y 2 RY : Then a probability like P(X > 0 and Y ≤ 0) is just X X p(x; y): x2RX :x>0 y2RY :y≤0 3 / 15 Joint Probability Distributions Two Discrete Random Variables ST 380 Probability and Statistics for the Physical Sciences Marginal Distribution To find the probability of an event defined only by X , we need the marginal pmf of X : X pX (x) = P(X = x) = p(x; y); x 2 RX : y2RY Similarly the marginal pmf of Y is X pY (y) = P(Y = y) = p(x; y); y 2 RY : x2RX 4 / 15 Joint
    [Show full text]
  • (Introduction to Probability at an Advanced Level) - All Lecture Notes
    Fall 2018 Statistics 201A (Introduction to Probability at an advanced level) - All Lecture Notes Aditya Guntuboyina August 15, 2020 Contents 0.1 Sample spaces, Events, Probability.................................5 0.2 Conditional Probability and Independence.............................6 0.3 Random Variables..........................................7 1 Random Variables, Expectation and Variance8 1.1 Expectations of Random Variables.................................9 1.2 Variance................................................ 10 2 Independence of Random Variables 11 3 Common Distributions 11 3.1 Ber(p) Distribution......................................... 11 3.2 Bin(n; p) Distribution........................................ 11 3.3 Poisson Distribution......................................... 12 4 Covariance, Correlation and Regression 14 5 Correlation and Regression 16 6 Back to Common Distributions 16 6.1 Geometric Distribution........................................ 16 6.2 Negative Binomial Distribution................................... 17 7 Continuous Distributions 17 7.1 Normal or Gaussian Distribution.................................. 17 1 7.2 Uniform Distribution......................................... 18 7.3 The Exponential Density...................................... 18 7.4 The Gamma Density......................................... 18 8 Variable Transformations 19 9 Distribution Functions and the Quantile Transform 20 10 Joint Densities 22 11 Joint Densities under Transformations 23 11.1 Detour to Convolutions......................................
    [Show full text]
  • Probability for Statistics: the Bare Minimum Prerequisites
    Probability for Statistics: The Bare Minimum Prerequisites This note seeks to cover the very bare minimum of topics that students should be familiar with before starting the Probability for Statistics module. Most of you will have studied classical and/or axiomatic probability before, but given the various backgrounds it's worth being concrete on what students should know going in. Real Analysis Students should be familiar with the main ideas from real analysis, and at the very least be familiar with notions of • Limits of sequences of real numbers; • Limit Inferior (lim inf) and Limit Superior (lim sup) and their connection to limits; • Convergence of sequences of real-valued vectors; • Functions; injectivity and surjectivity; continuous and discontinuous functions; continuity from the left and right; Convex functions. • The notion of open and closed subsets of real numbers. For those wanting for a good self-study textbook for any of these topics I would recommend: Mathematical Analysis, by Malik and Arora. Complex Analysis We will need some tools from complex analysis when we study characteristic functions and subsequently the central limit theorem. Students should be familiar with the following concepts: • Imaginary numbers, complex numbers, complex conjugates and properties. • De Moivre's theorem. • Limits and convergence of sequences of complex numbers. • Fourier transforms and inverse Fourier transforms For those wanting a good self-study textbook for these topics I recommend: Introduction to Complex Analysis by H. Priestley. Classical Probability and Combinatorics Many students will have already attended a course in classical probability theory beforehand, and will be very familiar with these basic concepts.
    [Show full text]
  • Using Learned Conditional Distributions As Edit Distance Jose Oncina, Marc Sebban
    Using Learned Conditional Distributions as Edit Distance Jose Oncina, Marc Sebban To cite this version: Jose Oncina, Marc Sebban. Using Learned Conditional Distributions as Edit Distance. Structural, Syntactic, and Statistical Pattern Recognition, Joint IAPR International Workshops, SSPR 2006 and SPR 2006, Aug 2006, Hong Kong, China. pp 403-411, ISBN 3-540-37236-9. hal-00322429 HAL Id: hal-00322429 https://hal.archives-ouvertes.fr/hal-00322429 Submitted on 18 Sep 2008 HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Using Learned Conditional Distributions as Edit Distance⋆ Jose Oncina1 and Marc Sebban2 1 Dep. de Lenguajes y Sistemas Informaticos,´ Universidad de Alicante (Spain) [email protected] 2 EURISE, Universite´ de Saint-Etienne, (France) [email protected] Abstract. In order to achieve pattern recognition tasks, we aim at learning an unbiased stochastic edit distance, in the form of a finite-state transducer, from a corpus of (input,output) pairs of strings. Contrary to the state of the art methods, we learn a transducer independently on the marginal probability distribution of the input strings. Such an unbiased way to proceed requires to optimize the pa- rameters of a conditional transducer instead of a joint one.
    [Show full text]
  • Multivariate Rank Statistics for Testing Randomness Concerning Some Marginal Distributions
    View metadata, citation and similar papers at core.ac.uk brought to you by CORE provided by Elsevier - Publisher Connector JOURNAL OF MULTIVARIATE ANALYSIS 5, 487-496 (1975) Multivariate Rank Statistics for Testing Randomness Concerning Some Marginal Distributions MARIE HuS~ovli Department of Statistics, CharZes University, Prague, Czechoskwahia Communicated by P. K. Sen In the present paper there is proposed a rank statistic for multivariate testing of randomness concerning some marginal distributions. The asymptotic distribu- tion of this statistic under hypothesis and “near” alternatives is treated. 1. INTRODUCTION Let Xj = (X, ,..., XJ, j = I,..., IV, be independent p-dimensional random variables and letF,&, y) andFdj(x) denote the distribution function of (Xii , XV!) and X,, , respectively. Fi, is assumedcontinuous throughout the whole paper, i=l ,..., p,j = l,...) N. We are interested in testing hypothesis of randomness of the univariate marginal distributions: for all x, i = l,..., p, j = 1,. N (1.1) against for all x, i = I,..., p, j = I ,..., N, (1.2) where $,, are unknown parameters. Or more generally, we are interested in testing hypothesis of randomnessassociated with some marginal distributions: F,“(x”) = F”(X”), u = l,..., Y, j = l,..., N, (1.3) against F,y(X”) = qx”; El;), Y = l,..., r, j = I,..., N, (1.4) Received September 1974; revised April 15, 1975. AMS subject classifications: Primary 62G10, Secondary 62HlO. Key words and phrases: Rank statistics; Testing randomness, Asymptotic distribution. 487 Copyright 0 1975 by Academic Press, Inc. All rights of reproduction in any form reserved. 488 MARIE HU?,KOVsi where FjV(xy) is the marginal distribution of the subvector Xy, Y = I,..., r, Xl ,***>x7, is a partition of the vector x, i.e., x’ = (xl’,..., xr’) and Oj’ = (0,” ,..., 0;‘) is a vector of unknown parameters.
    [Show full text]
  • Random Variables and Probability Distributions
    Schaum's Outline of Probability and Statistics CHAPTERCHAPTER 122 Random Variables and Probability Distributions Random Variables Suppose that to each point of a sample space we assign a number. We then have a function defined on the sam- ple space. This function is called a random variable (or stochastic variable) or more precisely a random func- tion (stochastic function). It is usually denoted by a capital letter such as X or Y. In general, a random variable has some specified physical, geometrical, or other significance. EXAMPLE 2.1 Suppose that a coin is tossed twice so that the sample space is S ϭ {HH, HT, TH, TT}. Let X represent the number of heads that can come up. With each sample point we can associate a number for X as shown in Table 2-1. Thus, for example, in the case of HH (i.e., 2 heads), X ϭ 2 while for TH (1 head), X ϭ 1. It follows that X is a random variable. Table 2-1 Sample Point HH HT TH TT X 2110 It should be noted that many other random variables could also be defined on this sample space, for example, the square of the number of heads or the number of heads minus the number of tails. A random variable that takes on a finite or countably infinite number of values (see page 4) is called a dis- crete random variable while one which takes on a noncountably infinite number of values is called a nondiscrete random variable. Discrete Probability Distributions Let X be a discrete random variable, and suppose that the possible values that it can assume are given by x1, x2, x3, .
    [Show full text]
  • Multivariate Distributions
    IEOR E4602: Quantitative Risk Management Spring 2016 c 2016 by Martin Haugh Multivariate Distributions We will study multivariate distributions in these notes, focusing1 in particular on multivariate normal, normal-mixture, spherical and elliptical distributions. In addition to studying their properties, we will also discuss techniques for simulating and, very briefly, estimating these distributions. Familiarity with these important classes of multivariate distributions is important for many aspects of risk management. We will defer the study of copulas until later in the course. 1 Preliminary Definitions Let X = (X1;:::Xn) be an n-dimensional vector of random variables. We have the following definitions and statements. > n Definition 1 (Joint CDF) For all x = (x1; : : : ; xn) 2 R , the joint cumulative distribution function (CDF) of X satisfies FX(x) = FX(x1; : : : ; xn) = P (X1 ≤ x1;:::;Xn ≤ xn): Definition 2 (Marginal CDF) For a fixed i, the marginal CDF of Xi satisfies FXi (xi) = FX(1;:::; 1; xi; 1;::: 1): It is straightforward to generalize the previous definition to joint marginal distributions. For example, the joint marginal distribution of Xi and Xj satisfies Fij(xi; xj) = FX(1;:::; 1; xi; 1;:::; 1; xj; 1;::: 1). If the joint CDF is absolutely continuous, then it has an associated probability density function (PDF) so that Z x1 Z xn FX(x1; : : : ; xn) = ··· f(u1; : : : ; un) du1 : : : dun: −∞ −∞ Similar statements also apply to the marginal CDF's. A collection of random variables is independent if the joint CDF (or PDF if it exists) can be factored into the product of the marginal CDFs (or PDFs). If > > X1 = (X1;:::;Xk) and X2 = (Xk+1;:::;Xn) is a partition of X then the conditional CDF satisfies FX2jX1 (x2jx1) = P (X2 ≤ x2jX1 = x1): If X has a PDF, f(·), then it satisfies Z xk+1 Z xn f(x1; : : : ; xk; uk+1; : : : ; un) FX2jX1 (x2jx1) = ··· duk+1 : : : dun −∞ −∞ fX1 (x1) where fX1 (·) is the joint marginal PDF of X1.
    [Show full text]
  • Comparing Conditional and Marginal Direct Estimation of Subgroup Distributions
    RESEARCH REPORT January 2003 RR-03-02 Comparing Conditional and Marginal Direct Estimation of Subgroup Distributions Matthias von Davier Research & Developm ent Division Princeton, NJ 08541 Comparing Conditional and Marginal Direct Estimation of Subgroup Distributions Matthias von Davier Educational Testing Service, Princeton, NJ January 2003 Research Reports provide preliminary and limited dissemination of ETS research prior to publication. They are available without charge from: Research Publications Office Mail Stop 10-R Educational Testing Service Princeton, NJ 08541 Abstract Many large-scale assessment programs in education utilize “conditioning models” that incorporate both cognitive item responses and additional respondent background variables relevant for the population of interest. The set of respondent background variables serves as a predictor for the latent traits (proficiencies/abilities) and is used to obtain a conditional prior distribution for these traits. This is done by estimating a linear regression, assuming normality of the conditional trait distributions given the set of background variables. Multiple imputations, or plausible values, of trait parameter estimates are used in addition to or, better, on top of the conditioning model—as a computationally convenient approach to generating consistent estimates of the trait distribution characteristics for subgroups in complex assessments. This report compares, on the basis of simulated and real data, the conditioning method with a recently proposed method of estimating subgroup distribution statistics that assumes marginal normality. Study I presents simulated data examples where the marginal normality assumption leads to a model that produces appropriate estimates only if subgroup differences are small. In the presence of larger subgroup differences that cannot be fitted by the marginal normality assumption, however, the proposed method produces subgroup mean and variance estimates that differ strongly from the true values.
    [Show full text]
  • Bivariate Kumaraswamy Models Via Modified Symmetric FGM Copulas: Properties and Applications in Insurance Modeling
    Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 25 September 2017 doi:10.20944/preprints201709.0115.v1 Peer-reviewed version available at J. Risk Financial Manag. 2017, 10, 19; doi:10.3390/jrfm10040019 Bivariate Kumaraswamy models via modified symmetric FGM copulas: Properties and Applications in Insurance modeling Indranil Ghosh Department of Mathematics and Statistics, University of North Carolina, Wilmington, USA Abstract A copula is a useful tool for constructing bivariate and/or multivariate dis- tributions. In this article, we consider a new modified class of (Farlie-Gumbel- Morgenstern) FGM bivariate copula for constructing several different bivariate Ku- maraswamy type copulas and discuss their structural properties, including depen- dence structures. It is established that construction of bivariate distributions by this method allows for greater flexibility in the values of Spearman's correlation coefficient, ρ and Kendall's τ. For illustrative purposes, one representative data set is utilized to exhibit the applicability of these proposed bivariate copula models. Key Words: Bivariate Kumaraswamy distribution, Copula based construction, Kendall's tau, Dependence structures 1 Introduction Over the last decade or so, there has been a growing interest in constructing various bivariate distributions and study it's dependence structure. For an excellent survey on this, an interested reader is suggested to see Balakrishnan et al. (2009) and the references therein. Of late, copula based methods of construction have also gained a considerable amount of attention, mainly due to it's analytical tractability in the sense of discussing 1 © 2017 by the author(s). Distributed under a Creative Commons CC BY license. Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 25 September 2017 doi:10.20944/preprints201709.0115.v1 Peer-reviewed version available at J.
    [Show full text]
  • Chapter 2 Random Variables and Distributions
    Chapter 2 Random Variables and Distributions CHAPTER OUTLINE Section 1 Random Variables Section 2 Distributions of Random Variables Section 3 Discrete Distributions Section 4 Continuous Distributions Section 5 Cumulative Distribution Functions Section 6 One-Dimensional Change of Variable Section 7 Joint Distributions Section 8 Conditioning and Independence Section 9 Multidimensional Change of Variable Section 10 Simulating Probability Distributions Section 11 Further Proofs (Advanced) In Chapter 1, we discussed the probability model as the central object of study in the theory of probability. This required defining a probability measure P on a class of subsets of the sample space S. It turns out that there are simpler ways of presenting a particular probability assignment than this — ways that are much more convenient to work with than P. This chapter is concerned with the definitions of random variables, distribution functions, probability functions, density functions, and the development of the concepts necessary for carrying out calculations for a probability model using these entities. This chapter also discusses the concept of the conditional distribution of one random variable, given the values of others. Conditional distributions of random variables provide the framework for discussing what it means to say that variables are related, which is important in many applications of probability and statistics. 33 34 Section 2.1: Random Variables 2.1 Random Variables The previous chapter explained how to construct probability models, including a sam- ple space S and a probability measure P. Once we have a probability model, we may define random variables for that probability model. Intuitively, a random variable assigns a numerical value to each possible outcome in the sample space.
    [Show full text]