Specification versus data fitting: SEM prediction and the Q‐class estimator

We propose a new class of limited information estimators built upon an explicit trade‐off between data fitting and a priori model specification. The estimators offer the researcher a continuum of estimators that range from an extreme emphasis on data fitting and robust reduced‐form estimation to the other extreme of exact model specification and efficient estimation. The approach used to generate the estimators illustrates why ULS often outperforms 2SLS‐PRRF even in the context of a correctly specified model, provides a new interpretation of 2SLS, and integrates Wonnacott and Wonnacott's (1970) least weighted variance estimators with other techniques. We apply the new class of estimators to Klein's Model I and generate forecasts. We find for this example that an emphasis on specification (as opposed to data fitting) produces better out‐of‐sample predictions. Copyright © 1999 John Wiley & Sons, Ltd.     

Predicting small region sectoral responses to changes in aggregate economic activity: A time series approach

The purpose of this paper is to investigate the applicability of a contemporary time series forecasting technique, transfer function modeling, to the problem of forecasting sectoral employment levels in small regional economies. The specific sectoral employment levels to be forecast are manufacturing, durable manufacturing, non-durable manufacturing and non-manufacturing employment. Due to data constraints at the small region level, construction of traditional causal econometric models is often very difficult; thus time series approaches become particularly attractive. The results suggest that transfer function models using readily available national indicator series as drivers can provide more accurate forecasts of small region sectoral employment levels than univariate time series models.     

Transfer and templates in scientific modelling

The notion of template has been advocated by Paul Humphreys and others as an illuminating unit of analysis in the philosophy of scientific modelling. Templates are supposed to have the dual functions of representing target systems and of facilitating quantitative manipulation. A resulting worry is that wide-ranging cross-disciplinary use of templates might compromise their representational function and reduce them to mere formalisms. In this paper, we argue that templates are valuable units of analysis in reconstructing cross-disciplinary modelling. Central to our discussion are the ways in which Lotka-Volterra models are used to analyse processes of technology diffusion. We illuminate both the similarities and differences between contributions to this case of cross-disciplinary modelling by reconstructing them as transfer of a template, without reducing the template to a mere formalism or a computational model. This requires differentiating the interpretation of templates from that of the models based on them. This differentiation allows us to claim that the LV models of technology diffusion that we review are the result of template transfer - conformist in some contributions, creative in others.     

基于模块性指标优化的层次聚类算法

层次形成的正确性决定了层次聚类的质量,通常围绕对象类内类间关系评价实现。本文基于聚类目标,综合考虑类内类问关系,借鉴网络分析中模块性评价准则,设计用于层次聚类的模块性指标,并采用自底向上合并的途径实现指标优化从而完成聚类,提出一种基于模块性指标优化的层次聚类算法。仿真试验表明,和谱聚类算法相比,本文介绍的算法实现简单,能以较少的计算代价,准确地获得样本特征,实现聚类。     

简单导联的非侵入便携式胎儿心电监护系统研究

提出并验证了一种简单导联方式实现的非侵入便携式胎儿心电监护仪的设计方案:以仪表放大器AD8220为核心,构建了母体胸导心电信号和腹壁混合信号的调理电路;通过AD7658实现A/D转换;然后将所得数字信号送入DSP芯片TMS320F28335,井通过USB通信模块CP2101将得到的心电信号数据传送到PC机。设计出了基于径向基函数的神经网络模块;提取出清晰的胎儿心电信号;调用绘图函数实现信号的实时显示。