多元成分数据的对数衬度偏最小二乘通径分析模型

本文研究多元成分数据的路径关联关系的建模问题,提出多元成分数据的对数衬度PLS通径分析模型.将中心化对数比变换与PLS通径分析方法相结合建立模型,其主要优势在于:①PLS通径分析模型对数据没有严格的分布假设要求,特别适于成分数据这类分布复杂的数据建模;②成分数据中心化对数比变换后的变量完全多重相关,PLS方法能够有效解决这一问题;③PLS通径分析模型特别适于多元成分数据这类具有层次关系的数据结构的建模,通过结构模型揭示多元成分数据之间的整体性路径关联关系,通过测量模型揭示成分数据与其成分分量之间的构成关系.更重要的是,本文的方法研究遵循成分数据所特有的代数基本理论,推导出模型的成分数据对数衬度隐变量的表达形式,从理论上证明了该建模方法的科学合理性.最后,将本方法用于北京市三次产业的投资结构、GDP结构、就业结构的路径关联关系的分析中,通过实证研究验证模型的可行性和应用价值.     

构成型顾客满意模型的偏最小二乘路径建模及其应用

本文研究了偏最小二乘路径建模在顾客满意模型中的应用,特别是引入了构成型关系的模型。本文首先比较了构成型模型和反映型模型的区别,并详尽阐述了构成型模型的偏最小二乘建模原理,接着构建了电信企业顾客满意度指数模型,并考虑了如何在指数模型中引入构成型外部关系.利用该电信企业的数据,比较分析了构成型模型(顾客期望和质量感知潜变量调整为构成型关系)和反映型模型(所有潜变量均为反映型关系)的实证结果,研究表明在为企业提供改善顾客满意水平的信息上两种模型具有较好的相似性,但是构成型模型能够提供更加稳定的结果,从而验证了顾客满意模型中引入构成型模型的可行性.     

On the convergence of the partial least squares path modeling algorithm

This paper adds to an important aspect of Partial Least Squares (PLS) path modeling, namely the convergence of the iterative PLS path modeling algorithm. Whilst conventional wisdom says that PLS always converges in practice, there is no formal proof for path models with more than two blocks of manifest variables. This paper presents six cases of non-convergence of the PLS path modeling algorithm. These cases were estimated using Mode A combined with the factorial scheme or the path weighting scheme, which are two popular options of the algorithm. As a conclusion, efforts to come to a proof of convergence under these schemes can be abandoned, and users of PLS should triangulate their estimation results.     

Non-symmetrical composite-based path modeling

Partial least squares path modeling presents some inconsistencies in terms of coherence with the predictive directions specified in the inner model (i.e. the path directions), because the directions of the links in the inner model are not taken into account in the iterative algorithm. In fact, the procedure amplifies interdependence among blocks and fails to distinguish between dependent and explanatory blocks. The method proposed in this paper takes into account and respects the specified path directions, with the aim of improving the predictive ability of the model and to maintain the hypothesized theoretical inner model. To highlight its properties, the proposed method is compared to the classical PLS path modeling in terms of explained variability, predictive relevance and interpretation using artificial data through a real data application. A further development of the method allows to treat multi-dimensional blocks in composite-based path modeling.     

Penalized least squares estimation with weakly dependent data

In statistics and machine learning communities, the last fifteen years have witnessed a surge of high-dimensional models backed by penalized methods and other state-of-the-art variable selection techniques. The high-dimensional models we refer to differ from conventional models in that the number of all parameters p and number of significant parameters s are both allowed to grow with the sample size T. When the field-specific knowledge is preliminary and in view of recent and potential affluence of data from genetics, finance and on-line social networks, etc., such (s, T, p)-triply diverging models enjoy ultimate flexibility in terms of modeling, and they can be used as a data-guided first step of investigation. However, model selection consistency and other theoretical properties were addressed only for independent data, leaving time series largely uncovered. On a simple linear regression model endowed with a weakly dependent sequence, this paper applies a penalized least squares (PLS) approach. Under regularity conditions, we show sign consistency, derive finite sample bound with high probability for estimation error, and prove that PLS estimate is consistent in L ₂ norm with rate \(\sqrt {s\log s/T}\).     

Variable selection via generalized SELO-penalized linear regression models

The seamless-L_0(SELO) penalty is a smooth function on [0, ∞) that very closely resembles the L_0 penalty, which has been demonstrated theoretically and practically to be effective in nonconvex penalization for variable selection. In this paper, we first generalize SELO to a class of penalties retaining good features of SELO, and then propose variable selection and estimation in linear models using the proposed generalized SELO(GSELO) penalized least squares(PLS) approach. We show that the GSELO-PLS procedure possesses the oracle property and consistently selects the true model under some regularity conditions in the presence of a diverging number of variables. The entire path of GSELO-PLS estimates can be efficiently computed through a smoothing quasi-Newton(SQN) method. A modified BIC coupled with a continuation strategy is developed to select the optimal tuning parameter. Simulation studies and analysis of a clinical data are carried out to evaluate the finite sample performance of the proposed method. In addition, numerical experiments involving simulation studies and analysis of a microarray data are also conducted for GSELO-PLS in the high-dimensional settings.     

An empirical study of PLAD regression using the bootstrap

Partial LAD regression uses the L ₁ norm associated with least absolute deviations (LAD) regression while retaining the same algorithmic structure of univariate partial least squares (PLS) regression. We use the bootstrap in order to assess the partial LAD regression model performance and to make comparisons to PLS regression. We use a variety of examples coming from NIR experiments as well as two sets of experimental data.     

Semiparametric and Additive Model Selection Using an Improved Akaike Information Criterion

Abstract

An improved AIC-based criterion is derived for model selection in general smoothing-based modeling, including semiparametric models and additive models. Examples are provided of applications to goodness-of-fit, smoothing parameter and variable selection in an additive model and semiparametric models, and variable selection in a model with a nonlinear function of linear terms.     

A Morse Index Theorem for Elliptic Operators on Bounded Domains

Given a selfadjoint, elliptic operator L, one would like to know how the spectrum changes as the spatial domain Ω ? ? ⁿ is deformed. For a family of domains {Ω _t } _{t∈[a, b]} we prove that the Morse index of L on Ω _a differs from the Morse index of L on Ω _b by the Maslov index of a path of Lagrangian subspaces on the boundary of Ω. This is particularly useful when Ω _a is a domain for which the Morse index is known, e.g. a region with very small volume. Then the Maslov index computes the difference of Morse indices for the “original” problem (on Ω _b ) and the “simplified” problem (on Ω _a ). This generalizes previous multi-dimensional Morse index theorems that were only available on star-shaped domains or for Dirichlet boundary conditions. We also discuss how one can compute the Maslov index using crossing forms, and present some applications to the spectral theory of Dirichlet and Neumann boundary value problems.     

A posteriori tests to validate dimension estimates from time series

A posteriori tests are proposed to evaluate the degree of reliability of the estimate of dimension from a time series, using the method of correlation integrals. Although we consider in particular the correlation dimension (D₂), our procedure can be applied to any generalized dimension as well as to the point-wise dimension.We propose the computation of two indices that can quantify to what degree the derivative of the log–log plot is constant with the correlation length and with the embedding dimension in the scaling region.An organized set of trials has been performed on time series from known model attractors, with different fractions of added measurement noise. On the basis of these trials, we found threshold values for the indices that discriminate between reliable and unreliable D₂ estimates.In the last part of the work we apply our procedure to real signals (electrocardiograms), finding good accordance between index values and the amount of noise in the time series.     

Image denoising via solution paths

In this paper, we propose a two-step kernel learning method based on the support vector regression (SVR) for financial time series forecasting. Given a number of candidate kernels, our method learns a sparse linear combination of these kernels so that the resulting kernel can be used to predict well on future data. The L ₁-norm regularization approach is used to achieve kernel learning. Since the regularization parameter must be carefully selected, to facilitate parameter tuning, we develop an efficient solution path algorithm that solves the optimal solutions for all possible values of the regularization parameter. Our kernel learning method has been applied to forecast the S&P500 and the NASDAQ market indices and showed promising results.     

Cpk index estimation using fuzzy numbers

Process capacity indices (PCIs) were developed and have been successfully used by companies to compete in and dominate the high-profit markets by improving the quality and the productivity since the past two decades. There is an essential assumption, in the conventional application, wherein the output process measurements are precise and distributed as normal random variables. Since the assumption of normal distribution is untenable, errors can occur if the C_pk index is computed using non-normal data. In the present study, we address the situation that the output of data from measurement of the quality of a product is insufficiently precise or scarce. This is possible when the quality measurement refers to the decision-maker’s subjective determination. In such a situation, the linguistic variable that is easier to capture the decision-maker’s subjective perception is applied to construct the PCI C_pk. The present approach can mitigate the effect when the normal assumption is inappropriate and extends the application of C_pk index.     

基于ELM算法的羊肉水分含量的快速无损检测

在近红外光谱900-1700nm的波长范围内采集南疆羊肉的光谱数据,来研究水分含量的快速无损检测.为减弱非目标因素对光谱的影响,采用SNV和去趋势法对光谱数据进行预处理.为降低建模的复杂度,去除共线性的影响,采用连续投影算法和相关系数法相结合选取8个特征波长变量,最后使用PLS和ELM算法分别进行建模.实验表明,与采用全光谱波段建模相比,采用特征波长变量建模,PLS和ELM算法的运行时间都大大缩短,并且在运行时间和预测精度上,ELM算法均优于PLS算法.ELM算法采用8个特征波段变量建模,预测精度达到0.9768,均方误差为4.4291e-04,相关系数为0.7603,运行时间可控制在1e-04s之下,这可为研发羊肉水分含量的便携式检测装置提供理论参考.     

On residuals in generalized Johnson SB regressions

An important challenge in the class of generalized Johnson S_B regression models is to define residuals which are capable of identifying departures from the model assumptions, as well as to assess the overall goodness-of-fit of the model. On this regard, we propose a new residual for this class of models, and numerically evaluate its behaviour relative to the deviance residual initially proposed for this class of regression models. Monte Carlo simulation experiments and empirical applications using real and simulated data are provided. Overall, the results favour the residual we propose.     

Ultrametricity indices for the Euclidean and Boolean hypercubes

Motivated by Murtagh’s experimental observation that sparse random samples of the hypercube become more and more ultrametric as the dimension increases, we consider a strict version of his ultrametricity coefficient, an index derived from Rammal’s degree of ultrametricity, and a topological ultrametricity index. First, we prove that the three ultrametricity indices converge in probability to one as dimension increases, if the sample size remains fixed. This is done for uniformly and normally distributed samples in the Euclidean hypercube, and for uniformly distributed samples in F₂ ^N with Hamming distance, as well as for very general probability distributions. Further, this holds true for random categorial data in complete disjunctive form. A second result is that the ultrametricity indices vanish in the limit for the full hypercube F₂ ^N as dimensionN increases,whereby Murtagh’s ultrametricity index is largest, and the topological ultrametricity index smallest, if N is large.     

Financial market forecasting using a two-step kernel learning method for the support vector regression

In this paper, we propose a two-step kernel learning method based on the support vector regression (SVR) for financial time series forecasting. Given a number of candidate kernels, our method learns a sparse linear combination of these kernels so that the resulting kernel can be used to predict well on future data. The L ₁-norm regularization approach is used to achieve kernel learning. Since the regularization parameter must be carefully selected, to facilitate parameter tuning, we develop an efficient solution path algorithm that solves the optimal solutions for all possible values of the regularization parameter. Our kernel learning method has been applied to forecast the S&P500 and the NASDAQ market indices and showed promising results.     

Study of Saudi Arabian climatic conditions using Hurst exponent and climatic predictability index

This paper utilizes Hurst exponent to study the persistency of meteorological parameters individually and dependency of rainfall/precipitation on pressure and temperature using climate predictability index. For the purpose, daily averages of surface pressure and temperature and daily total rainfall data for a period of 7 years for three locations and 14 years for seven locations has been utilized. The Hurst exponents (H) for above mentioned meteorological parameters were calculated using rescaled range analysis (R/S) and absolute moments methods. These H values were used to calculate the fractal dimension D for pressure, temperature and rainfall data. Finally, these D’s were used to calculate the climate predictability index PI_C in terms of pressure predictability index (PI_P), temperature predictability index (PI_T) and rainfall predictability index (PI_R). The Hurst exponent analysis showed that H values for rainfall, relative humidity and wind speed time series data for all the stations were >0.5 which is indicative of persistence behavior of the parameters on the previous values while for pressure and temperature H values were <0.5 means anti-persistence behavior. The climate predictability index showed that in most of the cases the rainfall was dependent on both pressure and temperature predictability indices. In some cases it was more dependent on pressure index than the temperature and in some cases otherwise. In Saudi Arabia there is no prevalent or established rainy season and the present analysis could not result into concrete results. It is therefore recommended to analyze the data by breaking the entire data set into seasons and further into different years.     

Stochastic binding of Ca2+ ions in the dyadic cleft; continuous vs random walk description of diffusion

Ca²⁺ signaling in the dyadic cleft in ventricular myocytes is fundamentally discrete and stochastic. In this paper we study the stochastic binding of single Ca²⁺ ions to receptors in the cleft using two different models of diffusion; a stochastic and discrete Random walk (RW) model, and a deterministic continuous model. We investigate if the latter model, together with a stochastic receptor model, can reproduce binding events registered in fully stochastic RW simulations. By evaluating the continuous model goodness-of-fit, we present evidences that it can. The large fluctuations in binding rate observed at the time level of single time steps are integrated and smoothed at the larger time scale of binding events, explaining the continuous model goodness-of-fit. With this we demonstrate that the stochasticity and discreteness of the Ca²⁺ signaling in the dyadic cleft, determined by single binding events, can be described with a deterministic model of Ca²⁺ diffusion together with a stochastic model of the binding events. Time consuming RW simulations can thus be avoided. We also present a new analytical model of bi-molecular binding probabilities that is used in the RW simulations, and in the statistical analysis. (© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)     

A Graphical Goodness-of-Fit Test for Dependence Models in Higher Dimensions

This article introduces a graphical goodness-of-fit test for copulas in more than two dimensions. The test is based on pairs of variables and can thus be interpreted as a first-order approximation of the underlying dependence structure. The idea is to first transform pairs of data columns with the Rosenblatt transform to bivariate standard uniform distributions under the null hypothesis. This hypothesis can be graphically tested with a matrix of bivariate scatterplots, Q-Q plots, or other transformations. Furthermore, additional information can be encoded as background color, such as measures of association or (approximate) p-values of tests of independence. The proposed goodness-of-fit test is designed as a basic graphical tool for detecting deviations from a postulated, possibly high-dimensional, dependence model. Various examples are given and the methodology is applied to a financial dataset. An implementation is provided by the R package copula. Supplementary material for this article is available online, which provides the R package copula and reproduces all the graphical results of this article.     

Extremal polyomino chains with respect to Zagreb indices

For a molecular graph, the first Zagreb index M₁ is equal to the sum of squares of the vertex degrees and second Zagreb index M₂ is equal to the sum of products of degree of pairs of adjacent vertices. In this paper, Zagreb indices of polyomino chains are computed. Also the extremal polyomino chains with respect to Zagreb indices are determined.