Rank-based empirical likelihood inference on medians of k populations

We propose a nonparametric method, called rank-based empirical likelihood (REL), for making inferences on medians and cumulative distribution functions (CDFs) of k populations. The standard distribution-free approach to testing the equality of k medians requires that the k population distributions have the same shape. Our REL-ratio (RELR) test for this problem requires fewer assumptions and can effectively use the symmetry information when the distributions are symmetric. Furthermore, our RELR statistic does not require estimation of variance, and achieves asymptotic pivotalness implicitly. When the k populations have equal medians we show that the REL method produces valid inferences for the common median and CDFs of k populations. Simulation results show that the REL approach works remarkably well in finite samples. A real data example is used to illustrate the proposed REL method.     

Quantifying the similarity of 2D images using edge pixels: an application to the forensic comparison of footwear impressions

We propose a novel method to quantify the similarity between an impression (Q) from an unknown source and a test impression (K) from a known source. Using the property of geometrical congruence in the impressions, the degree of correspondence is quantified using ideas from graph theory and maximum clique (MC). The algorithm uses the x and y coordinates of the edges in the images as the data. We focus on local areas in Q and the corresponding regions in K and extract features for comparison. Using pairs of images with known origin, we train a random forest to classify pairs into mates and non-mates. We collected impressions from 60 pairs of shoes of the same brand and model, worn over six months. Using a different set of very similar shoes, we evaluated the performance of the algorithm in terms of the accuracy with which it correctly classified images into source classes. Using classification error rates and ROC curves, we compare the proposed method to other algorithms in the literature and show that for these data, our method shows good classification performance relative to other methods. The algorithm can be implemented with the R package shoeprintr.     

Optimal selection and ordering of columns in supersaturated designs

Two methods to select columns for assigning factors to work on supersaturated designs are proposed. The focus of interest is the degree of non-orthogonality between the selected columns. One method is the exhaustive enumeration of selections of p columns from all k columns to find the exact optimality, while the other is intended to find an approximate solution by applying techniques used in the corresponding analysis, aiming for ease of use as well as a reduction in the large computing time required for large k with the first method. Numerical illustrations for several typical design matrices reveal that the resulting “approximately” optimal assignments of factors to their columns are exactly optimal for any p. Ordering the columns in E(s²)-optimal designs results in promising new findings including a large number of E(s²)-optimal designs.     

Two-tailed approximate confidence intervals for the ratio of proportions

Various approximate methods have been proposed for obtaining a two-tailed confidence interval for the ratio R of two proportions (independent samples). This paper evaluates 73 different methods (64 of which are new methods or modifications of older methods) and concludes that: (1) none of the classic methods (including the well-known score method) is acceptable since they are too liberal; (2), the best of the classic methods is the one based on logarithmic transformation (after increasing the data by 0.5), but it is only valid for large samples and moderate values of R; (3) the best methods among the 73 methods is based on an approximation to the score method (after adding 0.5 to all the data), with the added advantage of obtaining the interval by a simple method (i.e. solving a second degree equation); and (4) an option that is simpler than the previous one, and which is almost as effective for moderate values of R, consists of applying the classic Wald method (after adding a quantity to the data which is usually $z_{\alpha /2}^{2}/4$ ).     

Iterative numerical methods for sampling from high dimensional Gaussian distributions

Many applications require efficient sampling from Gaussian distributions. The method of choice depends on the dimension of the problem as well as the structure of the covariance- (Σ) or precision matrix (Q). The most common black-box routine for computing a sample is based on Cholesky factorization. In high dimensions, computing the Cholesky factor of Σ or Q may be prohibitive due to accumulation of more non-zero entries in the factor than is possible to store in memory. We compare different methods for computing the samples iteratively adapting ideas from numerical linear algebra. These methods assume that matrix vector products, Qv, are fast to compute. We show that some of the methods are competitive and faster than Cholesky sampling and that a parallel version of one method on a Graphical Processing Unit (GPU) using CUDA can introduce a speed-up of up to 30x. Moreover, one method is used to sample from the posterior distribution of petroleum reservoir parameters in a North Sea field, given seismic reflection data on a large 3D grid.     

Functional principal component analysis via regularized Gaussian basis expansions and its application to unbalanced data

This paper introduces regularized functional principal component analysis for multidimensional functional data sets, utilizing Gaussian basis functions. An essential point in a functional approach via basis expansions is the evaluation of the matrix for the integral of the product of any two bases (cross-product matrix). Advantages of the use of the Gaussian type of basis functions in the functional approach are that its cross-product matrix can be easily calculated, and it creates a much more flexible instrument for transforming each individual's observation into a functional form. The proposed method is applied to the analysis of three-dimensional (3D) protein structural data that can be referred to as unbalanced data. It is shown that our method extracts useful information from unbalanced data through the application. Numerical experiments are conducted to investigate the effectiveness of our method via Gaussian basis functions, compared to the method based on B-splines. On performing regularized functional principal component analysis with B-splines, we also derive the exact form of its cross-product matrix. The numerical results show that our methodology is superior to the method based on B-splines for unbalanced data.     

Improved likelihood-based inference for the stationary AR(2) model

An improved likelihood-based method based on Fraser et al. (1999) is proposed in this paper to test the significance of the second lag of the stationary AR(2) model. Compared with the test proposed by Fan and Yao (2003) and the signed log-likelihood ratio test, the proposed method has remarkable accuracy. Simulation studies are performed to illustrate the accuracy of the proposed method. Application of the proposed method on historical data is presented to demonstrate the implementation of this method. Furthermore, the method can be extended to the general AR(p) model.     

Estimation of quadratic functions of the Bernoulli parameter in inverse sampling method

The problem of estimating quadratic functions p² + βp (for different values of β), with reference to the estimation of genetic parameter and its functions, is discussed for the inverse sampling method. Then the inverse sampling method is compared with fixed size sampling method adopting the inverse of the precision per unit average sampling size as a criterion of comparison.     

A note on unbiasedness in ratio estimation

The method of ratio estimation for estimating the population mean ? of a characteristic y when we have auxillary information on a characteristic x highly correlated with y, consists in getting an estimator of the population ratio R = ?/X? and then multiplying this estimator by the known population mean X?. Though efficient, ratio estimators are in general biased and in this article we review some of the unbiased ratio estimators and discuss a method of constructing them. Next we present the Jackknife technique for reducing bias and show how the generalized Jackknife could be interpreted by the same method.     

Locally asymptotically optimal tests for AR(p) against diagonal bilinear dependence

The problem of testing linear AR(p1) against diagonal bilinear BL(p1, 0; p2, p2) dependence is considered. Emphasis is put on local asymptotic optimality and the nonspecification of innovation densities. The tests we are deriving are asymptotically valid under a large class of densities, and locally asymptotically most stringent at some selected density f. They rely on generalized versions of residual autocorrelations (the spectrum), and generalized versions of the so-called cubic autocorrelations (the bispectrum). Local powers are explicitly provided. The local power of the Gaussian Lagrange multipliers method follows as a particular case.     

A new characterization of Elfving's method for high dimensional computation

We give a new characterization of Elfving's (1952) method for computing c-optimal designs in k dimensions which gives explicit formulae for the k unknown optimal weights and k unknown signs in Elfving's characterization. This eliminates the need to search over these parameters to compute c-optimal designs, and thus reduces the computational burden from solving a family of optimization problems to solving a single optimization problem for the optimal finite support set. We give two illustrative examples: a high dimensional polynomial regression model and a logistic regression model, the latter showing that the method can be used for locally optimal designs in nonlinear models as well.     

Model selection with data-oriented penalty

We consider the problem of model (or variable) selection in the classical regression model using the GIC (general information criterion). In this method the maximum likelihood is used with a penalty function denoted by C_n, depending on the sample size n and chosen to ensure consistency in the selection of the true model. There are various choices of C_n suggested in the literature on model selection. In this paper we show that a particular choice of C_n based on observed data, which makes it random, preserves the consistency property and provides improved performance over a fixed choice of C_n.     

Interval estimation for the scale parameter of the two-parameter exponential distribution based on time-censored data

This paper is concerned with developing procedures for construcing confidence intervals, which would hold approximately equal tail probabilities and coverage probabilities close to the normal, for the scale parameter θ of the two-parameter exponential lifetime model when the data are time censored. We use a conditional approach to eliminate the nuisance parameter and develop several procedures based on the conditional likelihood. The methods are (a) a method based on the likelihood ratio, (b) a method based on the skewness corrected score (Bartlett, Biometrika 40 (1953), 12–19), (c) a method based on an adjustment to the signed root likelihood ratio (Diciccio, Field et al., Biometrika 77 (1990), 77–95), and (d) a method based on parameter transformation to the normal approximation. The performances of these procedures are then compared, through simulations, with the usual likelihood based procedure. The skewness corrected score procedure performs best in terms of holding both equal tail probabilities and nominal coverage probabilities even for small samples.     

Jackknifing two-sample statistics

In this paper, a new simple method for jackknifing two-sample statistics is proposed. The method is based on a two-step procedure. In the first step, the point estimator is calculated by leaving one X (or Y) out at a time. At the second step, the point estimator obtained in the first step is further jackknifed, leaving one Y (or X) out at a time, resulting in a simple formula for the proposed point estimator. It is shown that by using the two-step procedure, the bias of the point estimator is reduced in terms of asymptotic order, from O(n⁻¹) up to O(n⁻²), under certain regularity conditions. This conclusion is also confirmed empirically in terms of finite sample numerical examples via a small-scale simulation study. We also discuss the idea of asymptotic bias to obtain parallel results without imposing some conditions that may be difficult to check or too restrictive in practice.     

A comment on the use of auxiliary or surrogate outcome data

The method of double sampling in experiments has been recently rediscovered by Pepe (Biometrika 79 (1992), 355–365) and Pepe et al. (J. Statist. Plann. Inference 42 (1994), 137–160). However, a partially semi-parametric analysis method is employed, motivated by fears of misspecification of parts of the model and the belief that these parts are of little substantive interest. This paper points out the often substantial loss of information involved in the use of the method as compared with parametric analysis. It also argues that a broader review of the literature relevant to double sampling suggests that all of the model is usually of substantive interest and that the fears and consequences of misspecification are exaggerated.     

Designs in nonlinear regression by stochastic minimization of functionals of the mean square error matrix

We reconsider and extend the method of designing nonlinear experiments presented in Pázman and Pronzato (J. Statist. Plann. Inference 33 (1992) 385). The approach is based on the probability density of the LS estimators, and takes into account the boundary of the parameter space. The idea is to express the elements of the mean square error matrix (MSE) of the LS estimators as integrals of the density, express optimality criteria as functions of MSE, and minimize them by stochastic optimization. In the present paper we include prior knowledge about some parameters, we derive improved approximations of the density of estimators, and we consider not only linear optimality criteria like generalized A-optimality, but also D- and $L_s$-optimality criteria with an integer s. Of basic importance is the use of an accelerated method of stochastic optimization (MSO). This together with the important progress in computing allowed us to elaborate a realistic algorithm. Its performance is demonstrated by the search of a generalized D-optimum design in the 4-parameter growth curve model of microbiological experiments presented by Baranyi and Roberts (Internat. J. Food Microbiol. 23 (1994) 277; 26 (1995) 199).     

An alternative approach to the analysis of longitudinal data via generalized estimating equations

The generalized estimating equations (GEE) introduced by Liang and Zeger (Biometrika 73 (1986) 13–22) have been widely used over the past decade to analyze longitudinal data. The method uses a generalized quasi-score function estimate for the regression coefficients, and moment estimates for the correlation parameters. Recently, Crowder (Biometrika 82 (1995) 407–410) has pointed out some pitfalls with the estimation of the correlation parameters in the GEE method. In this paper we present a new method for estimating the correlation parameters which overcomes those pitfalls. For some commonly assumed correlation structures, we obtain unique feasible estimates for the correlation parameters. Large sample properties of our estimates are also established.     

Minimal orthogonal main-effect plans and partial replication

This work extends Jacroux's results (Jacroux, Technometrics 34 (1992) 92–96) involving minimal orthogonal main-effect plans (OMEP's). Tables of correct minimal numbers of runs of OMEP's are also provided. In addition, a general method is proposed to construct a class of minimal OMEP's. Moreover, a construction method is provided to obtain partially replicated OMEP's (PROMEP's). Our results provide more duplicate points for some plans than Jacroux's results on PROMEP's (Jacroux Technometrics 35 (1993) 32–36). In addition, a method is derived to construct s₁ × (s₁ − 1)²∥s²₁ minimal PROMEP's with maximal duplicate points.     

Construction of optimal supersaturated design with large number of levels

Inspired by the ideas of column and row juxtaposition in Liu and Lin (2009) and level transformation in Yamada and Lin (1999), this paper presents a new method for constructing optimal supersaturated designs (SSDs). This method provides a convenient way to construct mixed-level designs with relatively large numbers of levels, avoiding the blind search and numerous calculations by computers. The goodness of the resulting SSDs is judged by the χ² (Yamada and Lin, 1999 and Yamada and Matsui, 2002) and J₂ (Xu, 2002) criteria. Some nice properties of the new designs are also provided.     

Serial arrays and change-over designs

Consider change-over designs for the comparison of v treatments. Each application has a direct effect and up to n?1 residual and/or interaction effects. An array is defined to be a serial array if: (1) all treatments occur equally often in each row and (2) reading down columns, all possible permutations of n consecutive treatments occur equally often in the array. A method of constructing (λv+n?1) Xv^{n ?1}, λ integral, serial arrays is given. Use of these arrays as change-over designs is discussed with emphasis on the inclusion of interaction effects.