China
Africa
Explore chapters and articles related to this topic
CENTER: Critical Thinking in Team Decision-making
Published in Michael P. Letsky, Norman W. Warner, Stephen M. Fiore, C.A.P. Smith, Macrocognition in Teams, 2017
Kathleen P. Hess, Jared Freeman, Michael D. Coovert
The multilevel analyses began with a baseline or null model that served as a comparison against which to compare other substantive model solutions. Models were compared using the deviance statistic. Deviance reflects the overall lack of fit of the model to the data. Models with lower overall deviance terms are preferred, all other things being equal. Of course it is important not to add complexity to a model without ensuring the additional complexity significantly reduces overall deviance. This is easily examined as the deviance statistic for nested models is distributed as a chi-square. So the importance of the addition of a term to an equation (or other modification) can easily be checked by examining the decrease in overall deviance. The result of these multilevel analyses indicated that both individual-level variables (IQ and agreeableness) contributed significantly to team performance.
Automated Biventricular Cardiovascular Modelling from MRI for Big Heart Data Analysis
Published in Ervin Sejdić, Tiago H. Falk, Signal Processing and Machine Learning for Biomedical Big Data, 2018
Kathleen Gilbert, Xingyu Zhang, Beau Pontré, Avan Suinesiaputra, Pau Medrano-Gracia, Alistair Young
where P is the probability of a certain case belonging to the MI set, Xi are the values of the predictors, which in our case represent the PCA modes, βi are the coefficient terms of Xi and β0 is the intercept. The β terms were found by maximum likelihood estimation. The goodness of fit of the resulting model can be examined to determine how well the regression model distinguishes between non-patients and patients. Three common statistics used to quantify the goodness of fit of the model are deviance, Akaike information criterion (AIC) and Bayesian information criterion (BIC) [54]:
Employee Turnover Prediction Using Single Voting Model
Published in Pethuru Raj Chelliah, Usha Sakthivel, Nagarajan Susila, Applied Learning Algorithms for Intelligent IoT, 2021
R. Valarmathi, M. Umadevi, T. Sheela
Null deviance is the measure of the outcome predicted by the model without considering any input, whereas residual deviance is the measure of the outcome predicted by the model in consideration with the independent variable. Lower the value, better the performance.
Analysis of human-factor-caused freight train accidents in the United States
Published in Journal of Transportation Safety & Security, 2021
Zhipeng Zhang, Tejashree Turla, Xiang Liu
Using Equation 4, an NB model has been developed separately for the frequency of derailments caused due to human factors. Our model accounts for both traffic volume variable and year variable. Table 1 shows that the derailment rate, defined as the number of accidents normalized by traffic, is independent of traffic exposure (p > 0.05). However, as Equation 5 shows, accident frequency is linearly correlated with traffic volume given the same year. In addition, to test the correlation between train miles and year, a Pearson correlation test (Benesty et al., 2009) was used to test the correlation between train miles and year. Based upon the P-value (0.07, greater than the significance level alpha = 0.05), the test result indicates that there is no statistically significant correlation between the train mile and the year variable. However, the derailment frequency seems to have an annual decline of 6.1% from the parameter estimate of the year variable The model obtained from new parameters by excluding the traffic variable can be written as Equation 7. In order to evaluate the goodness of fit of these models, a statistical criterion called Deviance can be used. The Deviance approximately follows a chi-squared distribution and test checks if the null hypothesis of independence is true. The acceptable significance level is usually 5%. If the test statistic is improbably large, then the null hypothesis can be rejected, making it a good-fit.
Modelling fish physico-thermal habitat selection using functional regression
Published in Journal of Ecohydraulics, 2021
Jérémie Boudreault, André St-Hilaire, Fateh Chebana, Normand E. Bergeron
To assess models goodness-of-fit, the following ratio of deviance (D2) was calculated (Wood 2017), as commonly used in ecological studies (Guisan and Zimmermann 2000): where the null deviance is twice the difference between the log-likelihood of a saturated model (i.e. a model including exactly as many parameters as the number of observations), and the log-likelihood of the simplest possible model (i.e. a model containing only an intercept). The residual deviance is defined as twice the log-likelihood difference between the saturated model and the tested model and represents the deviance not explained by the model. The quantity D2 can be seen as an indicator of how close the tested model is to be perfect (D2 = 1) or to be the worst possible model (D2 = 0) (Guisan and Zimmermann 2000). When D2 is calculated for multiple linear regression models, it is equivalent to the traditional R2 (Neter et al. 1996).
The dose–response association between V̇O2peak and self-reported physical activity in children
Published in Journal of Sports Sciences, 2020
Alan M. Nevill, Michael J. Duncan, Gavin Sandercock
In order to provide a measure of goodness-of-fit “deviance” (−2 Log Likelihood) or more specifically the change in deviance was employed. The deviance statistic is a generalization of the sum of squares of residuals used in ordinary least squares but where model-fitting is achieved by maximum likelihood. It plays an important role in assessing the quality of fit in generalized linear models especially multilevel analyses. The difference between the deviances for competing models follows an approximate χ2 distribution with k-degrees of freedom (McCullagh & Nelder, 1989)