Effective use of screening experiments: Some practical experiences

When faced with a problem to be solved by running designed experiments we should first think of what strategy should be used for the experimentation. One aspect of successful experimental strategy is how to use screening experiments in practice. In this article the authors share their combined experiences of more than 90 years in using designed experiments in which screening experiments are an integral part of the strategy. Along the way the benefits of running screening experiments and why they work are discussed. Topics such as the Pareto principle applied to experimentation and the robustness of screening designs to assumptions are addressed. Case studies, tips, traps, and guidelines are provided. What emerges is a holistic approach to experimentation that results in practicing scientists and engineers getting the right data in the right amount at the right time.     

正交平衡区组设计统计模型中的方差分析

正交平衡区组设计(或广义正交表)的数据分析类似于正交拉丁方(或正交表)的数据分析.利用类似于正交表数据分析中的投影矩阵的正交分解技术,研究正交平衡区组设计的统计分析模型,给出了方差分析中的二次型以及各因子的二次型的分布性质,从而给出正交平衡区组设计统计模型中的方差分析方法.     

正交设计的最新发展和应用-回归分析在正交设计的应用

正交设计有许多新发展,本系列讲座介绍其中的一些便于应用的结果,共有四部份：回归分析在正交设计中的应用,均匀正交设计,正交设计的Ｄ－最优性,以及正交设计的投影性质。本讲强调回归分析用于正交设计的建模、估计、减少参数数目方面的应用     

正交平衡区组设计统计分析模型参数估计的矩阵表达

正交平衡区组设计(或者广义正交表)是一种类似于正交拉丁方(或者正交表)的新设计,但试验次数大幅减少.定义了一种基于正交相遇平衡区组设计(或者广义正交表)的统计分析模型,根据这个模型,给出了参数的最小二乘估计的矩阵形式.     

A Proof of the Jungnickel-Tonchev Conjecture on Quasi-Multiple Quasi-Symmetric Designs

A proof of the following conjecture of Jungnickel and Tonchev on quasi-multiple quasi-symmetric designs is given: Let D be a design whose parameter set (v,b,r,k,) equals (v,sv,sk,k, s) for some positive integer s and for some integers v,k, that satisfy (v-1) = k(k-1) (that is, these integers satisfy the parametric feasibility conditions for a symmetric (v,k,)-design). Further assume that D is a quasi-symmetric design, that is D has at most two block intersection numbers. If (k, (s-1)) = 1, then the only way D can be constructed is by taking multiple copies of a symmetric (v,k, )-design.     

正交平衡区组设计统计分析模型参数估计的分布特征研究

正交平衡区组设计(或者广义正交表)是一种类似于正交拉丁方(或者正交表)的新设计,但试验次数大幅减少.通过对正交平衡区组设计统计分析模型参数估计的分布特征进行了深入研究.研究发现,在试验数据正态性的情况下,各种参数估计也服从正态分布,并且各种参数的最小二乘估计都是无偏的,得到了各种参数估计的方差和独立性性质.     

单次重复试验的三水平析因设计的分析

由于试验材料、费用和时间等条件的限制,仅有单次重复试验的三水平析因设计经常要应用在农业、工业和医学临床试验等领域。例如,在医学临床试验中,为找到影响治疗关节炎效果的重要因子和最佳治疗方案需要考虑2个三水平的因子:A(药物治疗)和B(运动治疗),由于只能找到9位病情相似的病人进行试验,故只能实施仅有单次重复试验的三水平析因设计3~2。不幸的是,交互作用A×B也可能存在,这样就没有剩余自由度用于估计误差的方差,从而通常的方差分析方法不再能用于数据分析。针对上述问题,本文提出了三个基于均方误差的检验统计量用于分析单次重复试验的三水平析因设计。通过实例表明用这些方法不仅能检验所考虑因子的主效应,而且还能同时检验交互效应。相应检验所用的一些常用临界值提供在附录中。并且,还通过大量的模拟研究对所提出的三个检验方法进行了比较。结果显示,T_~((3))检验在三个检验方法中具有最大的功效。     

正交平衡区组设计统计分析模型的参数估计

正交平衡区组设计(或者广义正交表)的数据分析类似于正交拉丁方(或者正交表)的数据分析,但试验次数大幅减少.引入了相遇平衡区组设计矩阵象的概念,定义了一种基于正交相遇平衡区组设计(或者广义正交表)的统计分析模型,根据这个模型,推导得到了参数的最小二乘估计.     

A Note on the Construction of Orthogonal Latin Hypercube Designs

Latin hypercube designs have been found very useful for designing computer experiments. In recent years, several methods of constructing orthogonal Latin hypercube designs have been proposed in the literature. In this article, we report some more results on the construction of orthogonal Latin hypercubes which result in several new designs.     

一种构造多水平超饱和设计的新方法

超饱和设计在搜索试验中有重要应用.自上世纪九十年代迄今, 超饱和设计的研究取得了丰硕成果,研究的设计从二水平发展到多水平, 进而到混合水平;构造方法从巧妙的构思到利用组合理论进行系统构造,进而到各种算法的开发; 对于超饱和设计的评优准则也有更深入的认识.本文介绍一种构造多水平超饱和设计的新方法. 这种方法简单易行,很容易构造出具有优良性质的多水平超饱和设计, 有很强的实用性.     

固定设计下计算机模型L2校准的渐近性质

计算机仿真模型广泛应用于工业工程等领域,借助真实的物理观测对计算机模型的校准参数进行估计,可以提高计算机模型的仿真精度.已有文献指出,当物理试验的试验点独立同分布于均匀分布时,计算机模型校准参数的L2估计半参数有效.然而真实的物理试验点往往不满足独立同分布于均匀分布这一条件,文章证明了在固定的试验点下,较准参数L2估计的相合性和渐近正态性,并通过数值模拟验证了L2估计的收敛速度.     

8×4×2~n设计纯净两因子交互作用成分最大数的界

混水平部分因析设计在各类试验中有广泛应用.纯净效应准则是用于选取最优部分因析设计的重要准则之一.本文考虑含有一个八水平因子、一个四水平因子和若干二水平因子的8×4×2~n混水平设计,给出了分辨度为Ⅲ和Ⅳ的该类混水平设计包含纯净两因子交互作用成分最大数的上界和下界.下界通过构造特定设计而得到.     

On the Enumeration of ‐Optimal and Minimax‐Optimal k‐Circulant Supersaturated Designs

In recent years, several methods have been proposed for constructing ‐optimal and minimax‐optimal supersaturated designs (SSDs). However, until now the enumeration problem of such designs has not been yet considered. In this paper, ‐optimal and minimax‐optimal k‐circulant SSDs with 6, 10, 14, 18, 22, and 26 runs, factors and are enumerated in a computer search. We have also enumerated all ‐optimal and minimax‐optimal k‐circulant SSDs with (mod 4) and . The computer search utilizes the fact that theses designs are equivalent to certain 1‐rotational resolvable balanced incomplete block designs. Combinatorial properties of these resolvable designs are used to restrict the search space.     

A Localized Implementation of the Iterative Proportional Scaling Procedure for Gaussian Graphical Models

In this article, we propose localized implementations of the iterative proportional scaling (IPS) procedure by the strategy of partitioning cliques for computing maximum likelihood estimations in large Gaussian graphical models. We first divide the set of cliques into several nonoverlapping and nonempty blocks, and then adjust clique marginals in each block locally. Thus, high-order matrix operations can be avoided and the IPS procedure is accelerated. We modify the Swendsen–Wang Algorithm and apply the simulated annealing algorithm to find an approximation to the optimal partition which leads to the least complexity. This strategy of partitioning cliques can also speed up the existing IIPS and IHT procedures. Numerical experiments are presented to demonstrate the competitive performance of our new implementations and strategies.     

Bypassing the quadrature exactness assumption of hyperinterpolation on the sphere

This paper focuses on the approximation of continuous functions on the unit sphere by spherical polynomials of degree n via hyperinterpolation. Hyperinterpolation of degree n is a discrete approximation of the $L^2$-orthogonal projection of the same degree with its Fourier coefficients evaluated by a positive-weight quadrature rule that exactly integrates all spherical polynomials of degree at most 2n. This paper aims to bypass this quadrature exactness assumption by replacing it with the Marcinkiewicz–Zygmund property proposed in a previous paper. Consequently, hyperinterpolation can be constructed by a positive-weight quadrature rule (not necessarily with quadrature exactness). This scheme is referred to as unfettered hyperinterpolation. This paper provides a reasonable error estimate for unfettered hyperinterpolation. The error estimate generally consists of two terms: a term representing the error estimate of the original hyperinterpolation of full quadrature exactness and another introduced as compensation for the loss of exactness degrees. A guide to controlling the newly introduced term in practice is provided. In particular, if the quadrature points form a quasi-Monte Carlo (QMC) design, then there is a refined error estimate. Numerical experiments verify the error estimates and the practical guide.     

Asymptotically Optimal Tests and Optimal Designs for Testing the Mean in Regression Models with Applications to Change-Point Problems

Let a linear regression model be given with an experimental region [a, b] R and regression functions f ₁, ..., f _d+1 : [a, b] R. In practice it is an important question whether a certain regression function f _d+1, say, does or does not belong to the model. Therefore, we investigate the test problem H ₀ : "f _d+1 does not belong to the model" against K : "f _d+1 belong to the model" based on the least-squares residuals of the observations made at design points of the experimental region [a, b]. By a new functional central limit theorem given in Bischoff (1998, Ann. Statist. 26, 1398–1410), we are able to determine optimal tests in an asymptotic way. Moreover, we introduce the problem of experimental design for the optimal test statistics. Further, we compare the asymptotically optimal test with the likelihood ratio test (F-test) under the assumption that the error is normally distributed. Finally, we consider real change-point problems as examples and investigate by simulations the behavior of the asymptotic test for finite sample sizes. We determine optimal designs for these examples.     

非正则平移分布位置参数极大似然估计的极限分布及渐近有效性

In this paper, we consider the location model Y=θ e,where θ is an unknown parameter, and e is the error belonging to the interval [a, b]. We assume that e has the following density function.     

Guided Projections for Analyzing the Structure of High-Dimensional Data

A powerful data transformation method named guided projections is proposed creating new possibilities to reveal the group structure of high-dimensional data in the presence of noise variables. Using projections onto a space spanned by a selection of a small number of observations allows measuring the similarity of other observations to the selection based on orthogonal and score distances. Observations are iteratively exchanged from the selection creating a nonrandom sequence of projections, which we call guided projections. In contrast to conventional projection pursuit methods, which typically identify a low-dimensional projection revealing some interesting features contained in the data, guided projections generate a series of projections that serve as a basis not just for diagnostic plots but to directly investigate the group structure in data. Based on simulated data, we identify the strengths and limitations of guided projections in comparison to commonly employed data transformation methods. We further show the relevance of the transformation by applying it to real-world datasets.     

An upper bound of the number of tests in pooling designs for the error-tolerant complex model

Recently pooling designs have been used in screening experiments in molecular biology. In some applications, the property to be screened is defined on subsets of items, instead of on individual items. Such a model is usually referred to as the complex model. In this paper we give an upper bound of the number of tests required in a pooling design for the complex model (with given design parameters) where experimental errors are allowed.