P. McCullagh's research while affiliated with London Deanery and other places

Publications (30)

Citations

... Population response, defined as the difference between initial and final cell distribution (after a saturated-response experiment duration of 35 min), was measured five times for each species and culture light treatment. Differences in distribution were assessed using a log linear model (McCullagh and Nelder, 1983) constructed within a GenStat (9th Edition, VSN International Ltd., Hemel Hempstead, United Kingdom) program, and the proportion of cells displaying a preference for each light zone was tested using a three-model nested F-test (Hurley, 1996); to assess within treatment (hereafter termed: "F-test") and between treatment and species ("modified F-test") variation. Tukey tests of pairwise differences subsequently distinguished the specific zones in which significant accumulations or decreases in cell concentration occurred. ...
... When checking assumption for Poisson's regression model, it fulfills assumption of equi-dispersion. 15 Because of that Poisson's regression was used in this study. Finally, the odd ratios and the corresponding 95% confidence interval were calculated for each independent variable. ...
... The Jaccard Index is used (Goslee and Urban, 2007). Then, each dissimilarity matrix Second Kerguelen Plateau Symposium: marine ecosystem and fisheries Martin et al. (McCullagh and Nelder, 1989) is fitted, with the biological dissimilarity as the explained value and the environmental statistics as the explanatory factors. An iterative modelling process is run to select the best GLM and the best set of environmental factors, using the deviance explained as testing criteria (McCullagh and Nelder, 1989). ...
... Using CRF in named-entity problem [11], we denote = ( 1 , 2 , … , ) as the input sequence (words of a sentence) and = ( 1 , 2 , … , ) as the sequence of output states (named entity tags) and modeled the conditional probability as ...
... Species distribution modeling was performed using the presence records of species as the response variable (dependent variable) and 16 environmental variables as the predictor variables (independent variables) on the sdm R package (Naimi and Araujo 2016). In this study, 12 potential algorithms were tested, including maximum entropy (Maxent) (Phillips et al. 2006), generalized linear models (GLMs) (McCullagh and Nelder 1983), flexible discriminant analysis (FDA) (Hastie et al. 1994), boosted regression tree (BRT) (Friedman 2001), classification and regression tree (Cart) (Breiman et al. 1984), Glmnet (Friedman et al. 2010), multivariate adaptive regression splines (MARS) (Friedman 1991), maximum likelihood (Maxlike) (Royle et al. 2012), mixture discriminant analysis (MDA) (Hastie et al. 1994), recursive partitioning and regression trees (rpart) (Breiman et al. 1984), support vector machines (SVM) (Cortes and Vapnik 1995), and random forest (RF) (Breiman 2001) ( Table 2). In order to evaluate the efficiency of individual models and compare them with each other, the area under the curve (AUC) and correlation (Cor) statistics were used (Elith et al. 2006). ...
... However, in the iterative process to build joint models of the mean and dispersion there is some uncertainty in the estimation of φ i , thus Pinto and Pereira (2021) assume that φ i be replaced by τ φ i , where τ is an unknown constant. In this way, if H c and H d are two nested hypothesis of dimension c < d, that is η c ⊂ η d , then, considering the quasi-likelihood ratio statistic and according to McCullagh (1983), under H c the change in the extended quasi-deviance, given by ...
... The GLM, a typical nonlinear regression model, is an extension of the multiple linear regression model in linear statistical models [101]. GLM consist of a specific probability distribution from exponential family, a linear predictor, and a link function that relates the response variable to the linear predictor [102]. ...
... We adopted the 10% significance level in the preliminary step of including or excluding predictors in the surface response models. To determine the appropriate response surfaces for each response variable, predictors containing the highest p-value (p > 0.10) were progressively excluded respecting the hierarchy of effects: linear terms remained whenever interaction or quadratic terms were significant (MacCullagh and Nelder, 1983). In the final step of model fitting, we report the nominal significance levels (p-values) for the selected predictors. ...
... A variation on model (1) was used to examine the effect of flooding on the species richness of ants in pitfall traps. Given that species richness is likely to be Poisson distributed (i.e., small, non-negative integers), we used a generalized linear model suitable for Poisson-distributed variables (McCullagh and Nelder 1989): ...
... The ordinal logistic regression (OLR) [34,35] is a classification method for multiclass problems with a natural order among the response categories. Thus, it is perfectly suitable for the quality assessment experiment in which users issue scores within an ordered categorical scale (bad, poor, fair, good and excellent). ...