Bootstrap methods for bias correction and confidence interval estimation for nonlinear quantile regression of longitudinal data

This paper examines the use of bootstrapping for bias correction and calculation of confidence intervals (CIs) for a weighted nonlinear quantile regression estimator adjusted to the case of longitudinal data. Different weights and types of CIs are used and compared by computer simulation using a logistic growth function and error terms following an AR(1) model. The results indicate that bias correction reduces the bias of a point estimator but fails for CI calculations. A bootstrap percentile method and a normal approximation method perform well for two weights when used without bias correction. Taking both coverage and lengths of CIs into consideration, a non-bias-corrected percentile method with an unweighted estimator performs best.     

Testing for parameter constancy in the time series direction in panel data models

We propose tests for parameter constancy in the time series direction in panel data models. We construct a locally best invariant test based on Tanaka [Time series analysis: nonstationary and noninvertible distribution theory. New York: Wiley; 1996] and an asymptotically point optimal test based on Elliott and Müller [Efficient tests for general persistent time variation in regression coefficients. Rev Econ Stud. 2006;73:907–940]. We derive the limiting distributions of the test statistics as T→∞ while N is fixed, and calculate the critical values by applying numerical integration and response surface regression. Simulation results show that the proposed tests perform well if we apply them appropriately.     

基于多维效应的能源效率评估理论——以非参数DEA方法为例

基于多维效应的能源效率评估使人们在达到经济增长目的的同时兼顾能源安全、竞争力、环境、社会和技术的可持续性。论述能源效率内涵演化发展的路径，在此基础上将能源效率的内涵扩展至多维效应。以非参数DEA（Data Envelopment Analysis）方法为例，研究基于多维效应的能源效率评估方法的演化与发展趋势。研究结果显示:采用DEA方法对基于多维效应的能源效率进行评估的发展趋势主要有以下3点:能源效率的内涵不断延伸；DEA模型向非导向非径向模型转变；能源效率评估的研究维度不断扩展。对多维效应的能源效率综合指标评价体系和评价方法还需做深入研究。     

Opportunities and challenges of big data for the social sciences: The case of genomic data

In this paper, we draw attention to one unique and valuable source of big data, genomic data, by demonstrating the opportunities they provide to social scientists. We discuss different types of large-scale genomic data and recent advances in statistical methods and computational infrastructure used to address challenges in managing and analyzing such data. We highlight how these data and methods can be used to benefit social science research.     

The effectiveness of apprenticeship training: A within-track comparison of workplace-based and school-based vocational training in Hungary

Although apprenticeship training has been praised for its effectiveness in easing the transition of non-college-bound students from school to work, most studies rely on cross-country or cross-track comparisons. This study compares apprenticeship training students with non-apprentices within educational track in a relatively uncoordinated and decentralized institutional setting. Using a rich database and a unique set of observable individual-level characteristics as well as local labor market fixed effects to control for the potential selection bias, the results show that there are no significant differences in employment opportunities between apprentices and non-apprentices within just a year after graduation. This might be due to the failure of the Hungarian firms to enhance the skills of apprentices and thus increase their chances of entering the labor market compared to their school-trained peers. However, some immediate positive effect of apprenticeship training within sub-populations is apparent, which are likely to be the result of screening.     

Scanner Data and the Treatment of Quality Change in Nonrevisable Price Indexes

The recently developed rolling year GEKS procedure makes maximum use of all matches in the data to construct nonrevisable price indexes that are approximately free from chain drift. A potential weakness is that unmatched items are ignored. In this article we use imputation Törnqvist price indexes as inputs into the rolling year GEKS procedure. These indexes account for quality changes by imputing the “missing prices” associated with new and disappearing items. Three imputation methods are discussed. The first method makes explicit imputations using a hedonic regression model which is estimated for each time period. The other two methods make implicit imputations; they are based on time dummy hedonic and time-product dummy regression models and are estimated on bilateral pooled data. We present empirical evidence for New Zealand from scanner data on eight consumer electronics products and find that accounting for quality change can make a substantial difference.     

Multiple-class classification: Ordinal and categorical labels

We study multiple-class classification problems. Both ordinal and categorical labeled cases are discussed. The common approaches for multiple-class classification are built on binary classifiers, in which one-versus-one and one-versus-rest are typical approaches. When the number of classes is large, then these binary-classifier-based methods may suffer from either computational costs or the highly imbalanced sample sizes in their training stage. In order to alleviate the computational burden and the imbalanced training data issue in multiple-class classification problems, we propose a method that has competitive performance and retains the ease of model interpretation, which is essential for a prognostic/predictive model.     

Feature screening in ultrahigh-dimensional additive Cox model

The additive Cox model is flexible and powerful for modelling the dynamic changes of regression coefficients in the survival analysis. This paper is concerned with feature screening for the additive Cox model with ultrahigh-dimensional covariates. The proposed screening procedure can effectively identify active predictors. That is, with probability tending to one, the selected variable set includes the actual active predictors. In order to carry out the proposed procedure, we propose an effective algorithm and establish the ascent property of the proposed algorithm. We further prove that the proposed procedure possesses the sure screening property. Furthermore, we examine the finite sample performance of the proposed procedure via Monte Carlo simulations, and illustrate the proposed procedure by a real data example.     

A new algorithm for clustering based on kernel density estimation

In this paper, we present an algorithm for clustering based on univariate kernel density estimation, named ClusterKDE. It consists of an iterative procedure that in each step a new cluster is obtained by minimizing a smooth kernel function. Although in our applications we have used the univariate Gaussian kernel, any smooth kernel function can be used. The proposed algorithm has the advantage of not requiring a priori the number of cluster. Furthermore, the ClusterKDE algorithm is very simple, easy to implement, well-defined and stops in a finite number of steps, namely, it always converges independently of the initial point. We also illustrate our findings by numerical experiments which are obtained when our algorithm is implemented in the software Matlab and applied to practical applications. The results indicate that the ClusterKDE algorithm is competitive and fast when compared with the well-known Clusterdata and K-means algorithms, used by Matlab to clustering data.