			Home About us Contact

Good Predictive Power (good + predictive_power)

Distribution by Scientific Domains

Medical Sciences	33%
Chemistry	33%
Business, Economics, Finance and Accounting	33%

Selected Abstracts

Stacked partial least squares regression analysis for spectral calibration and prediction

JOURNAL OF CHEMOMETRICS, Issue 10 2009
Wangdong Ni
Abstract Two novel algorithms which employ the idea of stacked generalization or stacked regression, stacked partial least squares (SPLS) and stacked moving-window partial least squares (SMWPLS) are reported in the present paper. The new algorithms establish parallel, conventional PLS models based on all intervals of a set of spectra to take advantage of the information from the whole spectrum by incorporating parallel models in a way to emphasize intervals highly related to the target property. It is theoretically and experimentally illustrated that the predictive ability of these two stacked methods combining all subsets or intervals of the whole spectrum is never poorer than that of a PLS model based only on the best interval. These two stacking algorithms generate more parsimonious regression models with better predictive power than conventional PLS, and perform best when the spectral information is neither isolated to a single, small region, nor spread uniformly over the response. A simulation data set is employed in this work not only to demonstrate this improvement, but also to demonstrate that stacked regressions have the potential capability of predicting property information from an outlier spectrum in the prediction set. Moisture, oil, protein and starch in Cargill corn samples have been successfully predicted by these new algorithms, as well as hydroxyl number for different instruments of terpolymer samples including and excluding an outlier spectrum. Copyright © 2009 John Wiley & Sons, Ltd. [source]

Comparing leading theoretical models of behavioral predictions and post-behavior evaluations

PSYCHOLOGY & MARKETING, Issue 12 2008
Juliette Richetin
This study aimed at comparing the predictive power of the Theory of Planned Behavior (TPB), the Model of Goal-Directed Behavior (MGB), and the Extended Model of Goal-Directed Behavior (EMGB) for observed and self-reported behaviors concerning consumer nondurables. More specifically, the three models were compared in terms of their predictive power for intention and for behavioral desire (only MGB and EMGB). Additionally, the validity of four different models for predicting post-behavior evaluations was examined. Results showed that the EMGB is the most powerful in predicting both intention and behavioral desire. Moreover, results revealed that, as expected, all three models showed a better predictive power for SRB than for observed behavior. Finally, results demonstrated that post-behavior evaluations are both online and memory-based. ©2008 Wiley Periodicals, Inc. [source]

An Independent Evaluation of Four Quantitative Emergency Department Crowding Scales

ACADEMIC EMERGENCY MEDICINE, Issue 11 2006
Spencer S. Jones MStat
Background Emergency department (ED) overcrowding has become a frequent topic of investigation. Despite a significant body of research, there is no standard definition or measurement of ED crowding. Four quantitative scales for ED crowding have been proposed in the literature: the Real-time Emergency Analysis of Demand Indicators (READI), the Emergency Department Work Index (EDWIN), the National Emergency Department Overcrowding Study (NEDOCS) scale, and the Emergency Department Crowding Scale (EDCS). These four scales have yet to be independently evaluated and compared. Objectives The goals of this study were to formally compare four existing quantitative ED crowding scales by measuring their ability to detect instances of perceived ED crowding and to determine whether any of these scales provide a generalizable solution for measuring ED crowding. Methods Data were collected at two-hour intervals over 135 consecutive sampling instances. Physician and nurse agreement was assessed using weighted , statistics. The crowding scales were compared via correlation statistics and their ability to predict perceived instances of ED crowding. Sensitivity, specificity, and positive predictive values were calculated at site-specific cut points and at the recommended thresholds. Results All four of the crowding scales were significantly correlated, but their predictive abilities varied widely. NEDOCS had the highest area under the receiver operating characteristic curve (AROC) (0.92), while EDCS had the lowest (0.64). The recommended thresholds for the crowding scales were rarely exceeded; therefore, the scales were adjusted to site-specific cut points. At a site-specific cut point of 37.19, NEDOCS had the highest sensitivity (0.81), specificity (0.87), and positive predictive value (0.62). Conclusions At the study site, the suggested thresholds of the published crowding scales did not agree with providers' perceptions of ED crowding. Even after adjusting the scales to site-specific thresholds, a relatively low prevalence of ED crowding resulted in unacceptably low positive predictive values for each scale. These results indicate that these crowding scales lack scalability and do not perform as designed in EDs where crowding is not the norm. However, two of the crowding scales, EDWIN and NEDOCS, and one of the READI subscales, bed ratio, yielded good predictive power (AROC >0.80) of perceived ED crowding, suggesting that they could be used effectively after a period of site-specific calibration at EDs where crowding is a frequent occurrence. [source]

On currency crises and contagion

INTERNATIONAL JOURNAL OF FINANCE & ECONOMICS, Issue 2 2003
Marcel Fratzscher
Abstract This paper analyses the role of contagion in the currency crises in emerging markets during the 1990s. It employs a non-linear Markov-switching model to conduct a systematic comparison and evaluation of three distinct causes of currency crises: contagion, weak economic fundamentals, and sunspots, i.e. unobservable shifts in agents' beliefs. Testing this model empirically through Markov-switching and panel data models reveals that contagion, i.e. a high degree of real integration and financial interdependence among countries, is a core explanation for recent emerging market crises. The model has a remarkably good predictive power for the 1997,1998 Asian crisis. The findings suggest that in particular the degree of financial interdependence and also real integration among emerging markets are crucial not only in explaining past crises but also in predicting the transmission of future financial crises. Copyright © 2003 John Wiley & Sons, Ltd. [source]

Ab initio computational study of positron emission tomography ligands interacting with lipid molecule for the prediction of nonspecific binding

JOURNAL OF COMPUTATIONAL CHEMISTRY, Issue 14 2008
Lula Rosso
Abstract Nonspecific binding is a poorly understood biological phenomenon of relevance in the study of small molecules interactions in vivo and in drug development. Nonspecific binding is thought to be correlated in part to a molecule's lipophilicity, typically estimated by measuring (or calculating) octanol,water partition coefficient. This is, however, a gross simplification of a complex phenomenon. In this article, we present a computational method whose aim is to help identify positron emission tomography (PET) ligands with low nonspecific binding characteristics by investigating the molecular basis of ligand,membrane interaction. We considered a set consisting of 10 well-studied central nervous system PET radiotracers acting on a variety of molecular targets. Quantum mechanical calculations were used to estimate the strength of the interaction between each drug molecule and one phospholipid molecule commonly present in mammalian membranes. The results indicate a correlation between the computed drug,lipid interaction energy and the in vivo nonspecific distribution volume relative to the free tracer plasma concentration, calculated using standard compartmental modeling for the analysis of PET data. Significantly, the drugs whose interaction with the lipid molecule more favorably possessed, in general, a higher nonspecific binding value, whereas for the drugs taken in consideration in this study, the water-octanol partition coefficient, log P, did not show good predictive power of the nonspecific binding. This study also illustrates how ab initio chemical methods may offer meaningful and unbiased insights for the understanding of the underlying chemical mechanisms in biological systems. © 2008 Wiley Periodicals, Inc. J Comput Chem, 2008 [source]

Predicting post-operative delirium in elderly patients undergoing surgery for hip fracture

PSYCHOGERIATRICS, Issue 2 2006
Gregory GOLDENBERG
Abstract Background:, Delirium in elderly patients with hip fracture has a significant negative influence on the disease course. Awareness of risk factors for postoperative delirium (POD) may lead to the development of effective preventive strategies. The aims of this study were: to find patients' features that are predictors of POD, and; to develop a model predicting the risk for POD. Patients and methods:, Seventy-seven elderly patients (81.9 years of age, SD 7.5 years) were non-delirious prior to surgery and enrolled in the study. Delirium was diagnosed by Confusion Assessment Method and Algorrhithm. Patients' characteristics as potential predictors of POD were analyzed by logistic regression analysis on SAS software. Results:, Postoperative delirium was diagnosed in 37 patients. Use of multiple (>3) medications, lower scores on cognitive tests (<20 on Set Test and <24 on Mini-mental Status Exam), albumin level less than 3.5 g/dL, hematocrit level less than 33% and age over 81 years were predictors of POD. A logistic regression formula including these predictors weighed by their parameter estimates can be used to calculate the probability of POD. The model had a good fit and a good predictive power. A Delirium Predicting Scale was derived based on parameter estimates of these predictors. Patients can be classified as low-, intermediate- or high-risk for POD. Conclusions:, A logistic regression model, which includes patients' age, medication history, cognitive performance measured by Set Test and Mini-Mental Status Exam, albumin and hematocrit levels, can be used to predict risk for POD after surgical repair of fractured hip in elderly patients. [source]