DOCSLIB.ORG

Variance Function Estimation in High-Dimensions

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Variance Function Estimation in High-Dimensions Variance function estimation in high-dimensions Mladen Kolar [email protected] James Sharpnack [email protected] Machine Learning Department, Carnegie Mellon University, 5000 Forbes Ave, Pittsburgh, PA 15213 USA Abstract Penalization of the empirical loss by the `1 norm has become a popular tool for obtaining sparse models and We consider the high-dimensional het- huge amount of literature exists on theoretical prop- eroscedastic regression model, where the erties of estimation procedures (see,e.g., Zhao & Yu, mean and the log variance are modeled as 2006; Wainwright, 2009; Zhang, 2009; Zhang & Huang, a linear combination of input variables. Ex- 2008, and references therein) and on efficient algo- isting literature on high-dimensional linear rithms that numerically find estimates (see Bach et al., regression models has largely ignored non- 2011, for an extensive literature review). Due to limi- constant error variances, even though they tations of the ` norm penalization, high-dimensional commonly occur in a variety of applica- 1 inference methods based on the class of concave penal- tions ranging from biostatistics to finance. ties have been proposed that have better theoretical In this paper we study a class of non- and numerical properties (see,e.g., Fan & Li, 2001; Fan convex penalized pseudolikelihood estimators & Lv, 2009; Lv & Fan, 2009; Zhang & Zhang, 2011). for both the mean and variance parame- ters. We show that the Heteroscedastic Iter- In all of the above cited work, the main focus is on ative Penalized Pseudolikelihood Optimizer model selection and mean parameter estimation. Only (HIPPO) achieves the oracle property, that few papers deal with estimation of the variance in high- is, we prove that the rates of convergence are dimensions (Sun & Zhang, 2011; Fan et al., 2012) al- the same as if the true model was known. We though it is a fundamental problem in statistics. Vari- demonstrate numerical properties of the pro- ance appears in the confidence bounds on estimated cedure on a simulation study and real world regression coefficients and is important for variable se- data. lection as it appears in Akaike's information criterion (AIC) and the Bayesian information criterion (BIC). Furthermore, it provides confidence on the predictive 1. Introduction performance of a forecaster. High-dimensional regression models have been stud- In applied regression it is often the case that the er- ied extensively in both machine learning and statisti- ror variance is non-constant. Although the assumption cal literature. Statistical inference in high-dimensions, of a constant variance can sometimes be achieved by where the sample size n is smaller than the ambient transforming the dependent variable, e.g., by using a dimension p, is impossible without assumptions. As Box-Cox transformation, in many cases transforma- the concept of parsimony is important in many scien- tion does not produce a constant error variance (Car- tific domains, most of the research in the area of high- roll & Ruppert, 1988). Another approach is to ignore dimensional statistical inference is done under the as- the heterogeneous variance and use standard estima- sumption that the underlying model is sparse, in the tion techniques, but such estimators are less efficient. sense that the number of relevant parameters is much Aside from its use in reweighting schemes, estimat- smaller than p, or that it can be well approximated by ing variance is important because the resulting pre- a sparse model. diction intervals become more accurate and it is often important to explore which input variables drive the MK is partially supported through the grants NIH variance. In this paper, we will model the variance R01GM087694 and AFOSR FA9550010247. JS is partially supported by AFOSR under grant FA95501010382. directly as a parametric function of the explanatory variables. Appearing in Proceedings of the 29 th International Confer- ence on Machine Learning, Edinburgh, Scotland, UK, 2012. Heteroscedastic regression models are used in a vari- Copyright 2012 by the author(s)/owner(s). ety of fields ranging from biostatistics to economet- Variance function estimation in high-dimensions rics, finance and quality control in manufacturing. In common to assume that the support β is small, that this paper, we study penalized estimation in high- is, S = supp(β) and jSj n. Similarly, we assume dimensional heteroscedastic linear regression models, that the support T = supp(θ) is small. where the mean and the log variance are modeled as a linear combination of explanatory variables. Mod- 1.2. Related Work eling the log variance as a linear combination of the explanatory variables is a common choice as it guar- Consider the model (1) with constant variance, i.e., antees positivity and is also capable of capturing vari- σ(x; θ) ≡ σ0. Most of the existing high-dimensional ance that may vary over several orders of magnitudes literature is focused on estimation of the mean param- (Carroll & Ruppert, 1988; Harvey, 1976). Main contri- eter β in this homoscedastic regression model. Under butions of this paper are as follows. First, we propose a variety of assumptions and regularity conditions, any HIPPO (Heteroscedastic Iterative Penalized Pseudo- penalized estimation procedure mentioned in introduc- likelihood Optimizer) for estimation of both the mean tion can, in theory, select the correct sparse model with and variance parameters. Second, we establish the or- probability tending to 1. Literature on variance esti- acle property (in the sense of Fan & Lv (2009)) for mation is not as developed. Fan et al. (2012) proposed the estimated mean and variance parameters. Finally, a two step procedure for estimation of the unknown we demonstrate numerical properties of the proposed variance σ0, while (Sun & Zhang, 2011) proposed an procedure on a simulation study, where it is shown estimation procedure that jointly estimates the model that HIPPO outperforms other methods, and analyze and the variance. a real data set. Problem of estimation in the heteroscedastic linear re- gression models have been studied extensively in the 1.1. Problem Setup and Notation classical setting with p fixed, however, the problem of estimation under the model (1) when p n has not Consider the usual heteroscedastic linear model, been adequately studied. Jia et al. (2010) assume that 0 σ(x; θ) = jx0βj and show that Lasso is sign consistent yi = x β + σ(xi; θ)i; i = 1; : : : ; n; (1) i for the mean parameter β under certain conditions. 0 where X = (x1;:::; xn) = (X1;:::; Xp) is an n × p Their study shows limitations of lasso, for which many matrix of predictors with i.i.d. rows x1;:::; xn, y = highly scalable solvers exist. However, no new method- (y1; : : : ; yn) is an n-vector of responses, the vectors ology is developed, as the authors acknowledge that β 2 Rp and θ 2 Rp are p-vectors of mean and vari- the log-likelihood function is highly non-convex. Dette ance parameters, respectively, and = (1; : : : n) is & Wagener (2011) study the adaptive lasso under the an n-vector of i.i.d. random noise with mean 0 and model in (1). Under certain regularity conditions, they variance 1. We assume that the noise is indepen- show that the adaptive lasso is consistent, with sub- dent of the predictors X. The function σ(x; θ) has optimal asymptotic variance. However, the weighted a known parametric form and, for simplicity of pre- adaptive lasso is both consistent and achieves optimal sentation, we assume that it takes a particular form asymptotic variance, under the assumption that the 0 σ(xi; θ) = exp(xiθ=2). variance function is consistently estimated. However, they do not discuss how to obtain an estimator of the Throughout the paper we use [n] to denote the set variance function in a principled way and resort to f1; : : : ; ng. For any index set S ⊆ [p], we denote βS an ad-hoc fitting of the residuals. Daye et al. (2011) to be the subvector containing the components of the develop HHR procedure that optimizes the penalized vector β indexed by the set S, and XS denotes the log-likelihood under (1) with the ` -norm penalty on submatrix containing the columns of X indexed by 1 n both the mean and variance parameters. As the ob- S. For a vector a 2 R , we denote supp(a) = fj : jective is not convex, HHR estimates β with θ fixed aj 6= 0g the support set, jjajjq, q 2 (0; 1), the `q- and then estimates θ with β fixed, until convergence. norm defined as jjajj = (P aq)1=q with the usual q i2[n] i Since the objective is biconvex, HHR converges to a extensions for q 2 f0; 1g, that is, jjajj0 = jsupp(a)j stationary point. However, no theory is provided for and jjajj1 = maxi2[n] jaij. For notational simplicity, the final estimates. we denote jj · jj = jj · jj2 the `2 norm. For a matrix A 2 n×p we denote jjjAjjj the operator norm, jjAjj the R 2 F 2. Methodology Frobenius norm, and Λmin(A) and Λmax(A) denote the smallest and largest eigenvalue respectively. In this paper, we propose HIPPO (Heteroscedastic It- Under the model in (1), we are interested in estimating erative Penalized Pseudolikelihood Optimizer) for es- both β and θ. In high-dimensions, when p n, it is timating β and θ under model (1). Variance function estimation in high-dimensions In the first step, HIPPO finds the penalized pseudo- parameter with probability converging to one. Fur- likelihood maximizer of β by solving the following ob- thermore, on this support, the estimated coefficients jective should have the same asymptotic distribution as if an estimator that knew the true support was used. Such ^ 2 X β = arg min jjy − Xβjj + 2n ρλ (jβjj); (2) p S an estimator satisfies the oracle property.
Load more

Recommended publications
  • Generalized Linear Models and Generalized Additive Models
    00:34 Friday 27th February, 2015 Copyright ©Cosma Rohilla Shalizi; do not distribute without permission updates at http://www.stat.cmu.edu/~cshalizi/ADAfaEPoV/ Chapter 13 Generalized Linear Models and Generalized Additive Models [[TODO: Merge GLM/GAM and Logistic Regression chap- 13.1 Generalized Linear Models and Iterative Least Squares ters]] [[ATTN: Keep as separate Logistic regression is a particular instance of a broader kind of model, called a gener- chapter, or merge with logis- alized linear model (GLM). You are familiar, of course, from your regression class tic?]] with the idea of transforming the response variable, what we’ve been calling Y , and then predicting the transformed variable from X . This was not what we did in logis- tic regression. Rather, we transformed the conditional expected value, and made that a linear function of X . This seems odd, because it is odd, but it turns out to be useful. Let’s be specific. Our usual focus in regression modeling has been the condi- tional expectation function, r (x)=E[Y X = x]. In plain linear regression, we try | to approximate r (x) by β0 + x β. In logistic regression, r (x)=E[Y X = x] = · | Pr(Y = 1 X = x), and it is a transformation of r (x) which is linear. The usual nota- tion says | ⌘(x)=β0 + x β (13.1) · r (x) ⌘(x)=log (13.2) 1 r (x) − = g(r (x)) (13.3) defining the logistic link function by g(m)=log m/(1 m). The function ⌘(x) is called the linear predictor. − Now, the first impulse for estimating this model would be to apply the transfor- mation g to the response.
    [Show full text]
  • Stochastic Process - Introduction
    Stochastic Process - Introduction • Stochastic processes are processes that proceed randomly in time. • Rather than consider fixed random variables X, Y, etc. or even sequences of i.i.d random variables, we consider sequences X0, X1, X2, …. Where Xt represent some random quantity at time t. • In general, the value Xt might depend on the quantity Xt-1 at time t-1, or even the value Xs for other times s < t. • Example: simple random walk . week 2 1 Stochastic Process - Definition • A stochastic process is a family of time indexed random variables Xt where t belongs to an index set. Formal notation, { t : ∈ ItX } where I is an index set that is a subset of R. • Examples of index sets: 1) I = (-∞, ∞) or I = [0, ∞]. In this case Xt is a continuous time stochastic process. 2) I = {0, ±1, ±2, ….} or I = {0, 1, 2, …}. In this case Xt is a discrete time stochastic process. • We use uppercase letter {Xt } to describe the process. A time series, {xt } is a realization or sample function from a certain process. • We use information from a time series to estimate parameters and properties of process {Xt }. week 2 2 Probability Distribution of a Process • For any stochastic process with index set I, its probability distribution function is uniquely determined by its finite dimensional distributions. •The k dimensional distribution function of a process is defined by FXX x,..., ( 1 x,..., k ) = P( Xt ≤ 1 ,..., xt≤ X k) x t1 tk 1 k for anyt 1 ,..., t k ∈ I and any real numbers x1, …, xk .
    [Show full text]
  • Variance Function Regressions for Studying Inequality
    Variance Function Regressions for Studying Inequality The Harvard community has made this article openly available. Please share how this access benefits you. Your story matters Citation Western, Bruce and Deirdre Bloome. 2009. Variance function regressions for studying inequality. Working paper, Department of Sociology, Harvard University. Citable link http://nrs.harvard.edu/urn-3:HUL.InstRepos:2645469 Terms of Use This article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Open Access Policy Articles, as set forth at http:// nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of- use#OAP Variance Function Regressions for Studying Inequality Bruce Western1 Deirdre Bloome Harvard University January 2009 1Department of Sociology, 33 Kirkland Street, Cambridge MA 02138. E-mail: [email protected]. This research was supported by a grant from the Russell Sage Foundation. Abstract Regression-based studies of inequality model only between-group differ- ences, yet often these differences are far exceeded by residual inequality. Residual inequality is usually attributed to measurement error or the in- fluence of unobserved characteristics. We present a regression that in- cludes covariates for both the mean and variance of a dependent variable. In this model, the residual variance is treated as a target for analysis. In analyses of inequality, the residual variance might be interpreted as mea- suring risk or insecurity. Variance function regressions are illustrated in an analysis of panel data on earnings among released prisoners in the Na- tional Longitudinal Survey of Youth. We extend the model to a decomposi- tion analysis, relating the change in inequality to compositional changes in the population and changes in coefficients for the mean and variance.
    [Show full text]
  • Flexible Signal Denoising Via Flexible Empirical Bayes Shrinkage
    Journal of Machine Learning Research 22 (2021) 1-28 Submitted 1/19; Revised 9/20; Published 5/21 Flexible Signal Denoising via Flexible Empirical Bayes Shrinkage Zhengrong Xing [email protected] Department of Statistics University of Chicago Chicago, IL 60637, USA Peter Carbonetto [email protected] Research Computing Center and Department of Human Genetics University of Chicago Chicago, IL 60637, USA Matthew Stephens [email protected] Department of Statistics and Department of Human Genetics University of Chicago Chicago, IL 60637, USA Editor: Edo Airoldi Abstract Signal denoising—also known as non-parametric regression—is often performed through shrinkage estima- tion in a transformed (e.g., wavelet) domain; shrinkage in the transformed domain corresponds to smoothing in the original domain. A key question in such applications is how much to shrink, or, equivalently, how much to smooth. Empirical Bayes shrinkage methods provide an attractive solution to this problem; they use the data to estimate a distribution of underlying “effects,” hence automatically select an appropriate amount of shrinkage. However, most existing implementations of empirical Bayes shrinkage are less flexible than they could be—both in their assumptions on the underlying distribution of effects, and in their ability to han- dle heteroskedasticity—which limits their signal denoising applications. Here we address this by adopting a particularly flexible, stable and computationally convenient empirical Bayes shrinkage method and apply- ing it to several signal denoising problems. These applications include smoothing of Poisson data and het- eroskedastic Gaussian data. We show through empirical comparisons that the results are competitive with other methods, including both simple thresholding rules and purpose-built empirical Bayes procedures.
    [Show full text]
  • Variance Function Estimation in Multivariate Nonparametric Regression by T
    Variance Function Estimation in Multivariate Nonparametric Regression by T. Tony Cai, Lie Wang University of Pennsylvania Michael Levine Purdue University Technical Report #06-09 Department of Statistics Purdue University West Lafayette, IN USA October 2006 Variance Function Estimation in Multivariate Nonparametric Regression T. Tony Cail, Michael Levine* Lie Wangl October 3, 2006 Abstract Variance function estimation in multivariate nonparametric regression is consid- ered and the minimax rate of convergence is established. Our work uses the approach that generalizes the one used in Munk et al (2005) for the constant variance case. As is the case when the number of dimensions d = 1, and very much contrary to the common practice, it is often not desirable to base the estimator of the variance func- tion on the residuals from an optimal estimator of the mean. Instead it is desirable to use estimators of the mean with minimal bias. Another important conclusion is that the first order difference-based estimator that achieves minimax rate of convergence in one-dimensional case does not do the same in the high dimensional case. Instead, the optimal order of differences depends on the number of dimensions. Keywords: Minimax estimation, nonparametric regression, variance estimation. AMS 2000 Subject Classification: Primary: 62G08, 62G20. Department of Statistics, The Wharton School, University of Pennsylvania, Philadelphia, PA 19104. The research of Tony Cai was supported in part by NSF Grant DMS-0306576. 'Corresponding author. Address: 250 N. University Street, Purdue University, West Lafayette, IN 47907. Email: [email protected]. Phone: 765-496-7571. Fax: 765-494-0558 1 Introduction We consider the multivariate nonparametric regression problem where yi E R, xi E S = [0, Ild C Rd while a, are iid random variables with zero mean and unit variance and have bounded absolute fourth moments: E lail 5 p4 < m.
    [Show full text]
  • Variance Function Program
    Variance Function Program Version 15.0 (for Windows XP and later) July 2018 W.A. Sadler 71B Middleton Road Christchurch 8041 New Zealand Ph: +64 3 343 3808 e-mail: [email protected] (formerly at Nuclear Medicine Department, Christchurch Hospital) Contents Page Variance Function Program 15.0 1 Windows Vista through Windows 10 Issues 1 Program Help 1 Gestures 1 Program Structure 2 Data Type 3 Automation 3 Program Data Area 3 User Preferences File 4 Auto-processing File 4 Graph Templates 4 Decimal Separator 4 Screen Settings 4 Scrolling through Summaries and Displays 4 The Variance Function 5 Variance Function Output Examples 8 Variance Stabilisation 11 Histogram, Error and Biological Variation Plots 12 Regression Analysis 13 Limit of Blank (LoB) and Limit of Detection (LoD) 14 Bland-Altman Analysis 14 Appendix A (Program Delivery) 16 Appendix B (Program Installation) 16 Appendix C (Program Removal) 16 Appendix D (Example Data, SD and Regression Files) 16 Appendix E (Auto-processing Example Files) 17 Appendix F (History) 17 Appendix G (Changes: Version 14.0 to Version 15.0) 18 Appendix H (Version 14.0 Bug Fixes) 19 References 20 Variance Function Program 15.0 1 Variance Function Program 15.0 The variance function (the relationship between variance and the mean) has several applications in statistical analysis of medical laboratory data (eg. Ref. 1), but the single most important use is probably the construction of (im)precision profile plots (2). This program (VFP.exe) was initially aimed at immunoassay data which exhibit relatively extreme variance relationships. It was based around the “standard” 3-parameter variance function, 2 J σ = (β1 + β2U) 2 where σ denotes variance, U denotes the mean, and β1, β2 and J are the parameters (3, 4).
    [Show full text]
  • Examining Residuals
    Stat 565 Examining Residuals Jan 14 2016 Charlotte Wickham stat565.cwick.co.nz So far... xt = mt + st + zt Variable Trend Seasonality Noise measured at time t Estimate and{ subtract off Should be left with this, stationary but probably has serial correlation Residuals in Corvallis temperature series Temp - loess smooth on day of year - loess smooth on date Your turn Now I have residuals, how could I check the variance doesn't change through time (i.e. is stationary)? Is the variance stationary? Same checks as for mean except using squared residuals or absolute value of residuals. Why? var(x) = 1/n ∑ ( x - μ)2 Converts a visual comparison of spread to a visual comparison of mean. Plot squared residuals against time qplot(date, residual^2, data = corv) + geom_smooth(method = "loess") Plot squared residuals against season qplot(yday, residual^2, data = corv) + geom_smooth(method = "loess", size = 1) Fitted values against residuals Looking for mean-variance relationship qplot(temp - residual, residual^2, data = corv) + geom_smooth(method = "loess", size = 1) Non-stationary variance Just like the mean you can attempt to remove the non-stationarity in variance. However, to remove non-stationary variance you divide by an estimate of the standard deviation. Your turn For the temperature series, serial dependence (a.k.a autocorrelation) means that today's residual is dependent on yesterday's residual. Any ideas of how we could check that? Is there autocorrelation in the residuals? > corv$residual_lag1 <- c(NA, corv$residual[-nrow(corv)]) > head(corv) . residual residual_lag1 . 1.5856663 NA xt-1 = lag 1 of xt . -0.4928295 1.5856663 .
    [Show full text]
  • Estimating Variance Functions for Weighted Linear Regression
    Kansas State University Libraries New Prairie Press Conference on Applied Statistics in Agriculture 1992 - 4th Annual Conference Proceedings ESTIMATING VARIANCE FUNCTIONS FOR WEIGHTED LINEAR REGRESSION Michael S. Williams Hans T. Schreuder Timothy G. Gregoire William A. Bechtold See next page for additional authors Follow this and additional works at: https://newprairiepress.org/agstatconference Part of the Agriculture Commons, and the Applied Statistics Commons This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License. Recommended Citation Williams, Michael S.; Schreuder, Hans T.; Gregoire, Timothy G.; and Bechtold, William A. (1992). "ESTIMATING VARIANCE FUNCTIONS FOR WEIGHTED LINEAR REGRESSION," Conference on Applied Statistics in Agriculture. https://doi.org/10.4148/2475-7772.1401 This is brought to you for free and open access by the Conferences at New Prairie Press. It has been accepted for inclusion in Conference on Applied Statistics in Agriculture by an authorized administrator of New Prairie Press. For more information, please contact [email protected]. Author Information Michael S. Williams, Hans T. Schreuder, Timothy G. Gregoire, and William A. Bechtold This is available at New Prairie Press: https://newprairiepress.org/agstatconference/1992/proceedings/13 Conference on153 Applied Statistics in Agriculture Kansas State University ESTIMATING VARIANCE FUNCTIONS FOR WEIGHTED LINEAR REGRESSION Michael S. Williams Hans T. Schreuder Multiresource Inventory Techniques Rocky Mountain Forest and Range Experiment Station USDA Forest Service Fort Collins, Colorado 80526-2098, U.S.A. Timothy G. Gregoire Department of Forestry Virginia Polytechnic Institute and State University Blacksburg, Virginia 24061-0324, U.S.A. William A. Bechtold Research Forester Southeastern Forest Experiment Station Forest Inventory and Analysis Project USDA Forest Service Asheville, North Carolina 28802-2680, U.S.A.
    [Show full text]
  • Variance Function Regressions for Studying Inequality
    Variance Function Regressions for Studying Inequality The Harvard community has made this article openly available. Please share how this access benefits you. Your story matters Citation Western, Bruce and Deirdre Bloome. 2009. Variance function regressions for studying inequality. Working paper, Department of Sociology, Harvard University. Citable link http://nrs.harvard.edu/urn-3:HUL.InstRepos:2645469 Terms of Use This article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Open Access Policy Articles, as set forth at http:// nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of- use#OAP Variance Function Regressions for Studying Inequality Bruce Western1 Deirdre Bloome Harvard University January 2009 1Department of Sociology, 33 Kirkland Street, Cambridge MA 02138. E-mail: [email protected]. This research was supported by a grant from the Russell Sage Foundation. Abstract Regression-based studies of inequality model only between-group differ- ences, yet often these differences are far exceeded by residual inequality. Residual inequality is usually attributed to measurement error or the in- fluence of unobserved characteristics. We present a regression that in- cludes covariates for both the mean and variance of a dependent variable. In this model, the residual variance is treated as a target for analysis. In analyses of inequality, the residual variance might be interpreted as mea- suring risk or insecurity. Variance function regressions are illustrated in an analysis of panel data on earnings among released prisoners in the Na- tional Longitudinal Survey of Youth. We extend the model to a decomposi- tion analysis, relating the change in inequality to compositional changes in the population and changes in coefficients for the mean and variance.
    [Show full text]
  • Profiling Heteroscedasticity in Linear Regression Models
    358 The Canadian Journal of Statistics Vol. 43, No. 3, 2015, Pages 358–377 La revue canadienne de statistique Profiling heteroscedasticity in linear regression models Qian M. ZHOU1*, Peter X.-K. SONG2 and Mary E. THOMPSON3 1Department of Statistics and Actuarial Science, Simon Fraser University, Burnaby, British Columbia, Canada 2Department of Biostatistics, University of Michigan, Ann Arbor, Michigan, U.S.A. 3Department of Statistics and Actuarial Science, University of Waterloo, Waterloo, Ontario, Canada Key words and phrases: Heteroscedasticity; hybrid test; information ratio; linear regression models; model- based estimators; sandwich estimators; screening; weighted least squares. MSC 2010: Primary 62J05; secondary 62J20 Abstract: Diagnostics for heteroscedasticity in linear regression models have been intensively investigated in the literature. However, limited attention has been paid on how to identify covariates associated with heteroscedastic error variances. This problem is critical in correctly modelling the variance structure in weighted least squares estimation, which leads to improved estimation efficiency. We propose covariate- specific statistics based on information ratios formed as comparisons between the model-based and sandwich variance estimators. A two-step diagnostic procedure is established, first to detect heteroscedasticity in error variances, and then to identify covariates the error variance structure might depend on. This proposed method is generalized to accommodate practical complications, such as when covariates associated with the heteroscedastic variances might not be associated with the mean structure of the response variable, or when strong correlation is present amongst covariates. The performance of the proposed method is assessed via a simulation study and is illustrated through a data analysis in which we show the importance of correct identification of covariates associated with the variance structure in estimation and inference.
    [Show full text]
  • Variance Function Regressions for Studying Inequality
    9 VARIANCE FUNCTION REGRESSIONS FOR STUDYING INEQUALITY Bruce Western* Deirdre Bloome* Regression-based studies of inequality model only between-group differences, yet often these differences are far exceeded by resid- ual inequality. Residual inequality is usually attributed to mea- surement error or the influence of unobserved characteristics. We present a model, called variance function regression,thatin- cludes covariates for both the mean and variance of a dependent variable. In this model, the residual variance is treated as a target for analysis. In analyses of inequality, the residual variance might be interpreted as measuring risk or insecurity. Variance function regressions are illustrated in an analysis of panel data on earnings among released prisoners in the National Longitudinal Survey of Youth. We extend the model to a decomposition analysis, relating the change in inequality to compositional changes in the popula- tion and changes in coefficients for the mean and variance. The decomposition is applied to the trend in U.S. earnings inequality among male workers, 1970 to 2005. 1. INTRODUCTION In studying inequality, we can distinguish differences between groups from differences within groups. Sociological theory usually motivates This research was supported by a grant from the Russell Sage Foundation. Direct correspondence to Bruce Western, Department of Sociology, 33 Kirkland Street, Cambridge MA 02138; e-mail: [email protected]. *Harvard University 293 294 WESTERN AND BLOOME hypotheses about between-group inequality. For these hypotheses, in- terest focuses on differences in group averages. For example, theories of labor market discrimination predict whites earn more than blacks, and men earn more than women. Human capital theory explains why college graduates average higher earnings than high school dropouts.
    [Show full text]
  • Wavelet-Based Functional Mixed Models
    Wavelet-Based Functional Mixed Models Jeffrey S. Morris 1 The University of Texas MD Anderson Cancer Center, Houston, USA and Raymond J. Carroll Texas A&M University, College Station, USA Summary: Increasingly, scientific studies yield functional data, in which the ideal units of observation are curves and the observed data consist of sets of curves sampled on a fine grid. In this paper, we present new methodology that generalizes the linear mixed model to the functional mixed model framework, with model fitting done using a Bayesian wavelet-based approach. This method is flexible, allowing functions of arbitrary form and the full range of fixed effects structures and between-curve covariance structures available in the mixed model framework. It yields nonparametric estimates of the fixed and random effects functions as well as the various between-curve and within-curve covariance matrices. The functional fixed effects are adaptively regularized as a result of the nonlinear shrinkage prior imposed on the fixed effects’ wavelet coefficients, and the random effect functions experience a form of adaptively regularization because of the separately estimated variance components for each wavelet coefficient. Because we have posterior samples for all model quantities, we can perform pointwise or joint Bayesian inference or prediction on the quantities of the model. The adaptiveness of the method makes it especially appropriate for modeling irregular functional data characterized by numerous local features like peaks. Keywords: Bayesian methods; Functional data analysis; Mixed models; Model averaging; Non- parametric regression; Proteomics; Wavelets Short title: Wavelet-Based Functional Mixed Models 1 Introduction Technological innovations in science and medicine have resulted in a growing number of scientific studies that yield functional data.
    [Show full text]