Computing Confidence Set for the Kullback-Leibler Best Model
This function computes the confidence set on the best model given the data and model set. confset implements three different methods proposed by Burnham and Anderson (2002).
1.1
confset(cand.set, modnames = NULL, second.ord = TRUE, nobs = NULL, method = "raw", level = 0.95, delta = 6, c.hat = 1)
cand.set: a list storing each of the models in the candidate model set.
modnames: a character vector of model names to facilitate the identification of each model in the model selection table. If NULL, the function uses the names in the cand.set list of candidate models. If no names appear in the list, generic names (e.g., Mod1, Mod2) are supplied in the table in the same order as in the list of candidate models.
second.ord: logical. If TRUE, the function returns the second-order Akaike information criterion (i.e., AICc).
nobs: this argument allows to specify a numeric value other than total sample size to compute the AICc (i.e., nobs defaults to total number of observations). This is relevant only for mixed models or various models of unmarkedFit classes where sample size is not straightforward. In such cases, one might use total number of observations or number of independent clusters (e.g., sites) as the value of nobs.
method: a character value, either as raw, ordinal, or ratio, indicating the method for determining the confidence set for the best model (see 'Description' above for details).
level: the level of confidence (i.e., sum of model probabilities) used to determine the confidence set on the best model when using the raw
method. Note that the argument is not used for the other methods of determining the confidence set on the best model.
delta: the delta (Q)AIC(c) value associated with the cutoff point to determine the confidence set for the best model. Note that the argument is only used when method = ratio.
c.hat: value of overdispersion parameter (i.e., variance inflation factor) such as that obtained from c_hat. Note that values of c.hat different from 1 are only appropriate for binomial GLM's with trials > 1 (i.e., success/trial or cbind(success, failure) syntax), with Poisson GLM's, single-season occupancy models (MacKenzie et al. 2002), dynamic occupancy models (MacKenzie et al. 2003), or N-mixture models (Royle 2004, Dail and Madsen 2011). If c.hat > 1, confset will return the quasi-likelihood analogue of the information criteria requested and multiply the variance-covariance matrix of the estimates by this value (i.e., SE's are multiplied by sqrt(c.hat)). This option is not supported for generalized linear mixed models of the mer or merMod classes.
The first and simplest (method = 'raw'), relies on summing the Akaike weights (i.e., model probabilities) of the ranked models until we reach a given cutpoint (e.g., 0.95 for a 95 percent set).
The second method (method = 'ordinal') suggested is based on the classification of the models on an ordinal scale based on the delta (Q)AIC(c). The models are grouped in different classes based on their weight of support as determined by the delta (Q)AIC(c) values: substantial support (delta (Q)AIC(c) <= 2), some support (2 < delta (Q)AIC(c) <= 7), little support (7 < delta (Q)AIC(c) <= 10), no support (delta (Q)AIC(c) > 10).
The third method (method = 'ratio') is based on identifying the ratios of model likelihoods (i.e., exp(-delta_(Q)AIC(c)/2) ) that exceed a cutpoint, similar to the building of profile likelihood intervals. An evidence ratio of each model relative to the top-ranked model is computed and the ratios exceeding the cutpoint determine which models are included in the confidence set. Note here that small cutoff points are suggested (e.g., 0.125, 0.050). The cutoff point is linked to delta (Q)AIC(c) by the following relationship: c("", "").
confset returns an object of class confset as a list with the following components, depending on which method is used:
when method = 'raw': - method: identifies the method of determining the confidence set on the best model.
level: the confidence level used to determine the confidence set on the best model.
table: a reduced table with the models included in the confidence set.
when method = 'ordinal': - method: identifies the method of determining the confidence set on the best model.
substantial: a reduced table with the models included in the confidence set for which delta (Q)AIC(c) <= 2.
some: a reduced table with the models included in the confidence set for which 2 < delta (Q)AIC(c) <= 7.
little: a reduced table with the models included in the confidence set for which 7 < delta (Q)AIC(c) <= 10.
none: a reduced table with the models included in the confidence set for which delta (Q)AIC(c) > 10.
when method = 'ratio': - method: identifies the method of determining the confidence set on the best model.
cutoff: the cutoff value for the ratios used to determine the confidence set on the best model.
delta: the delta (Q)AIC(c) used to compute the cutoff value for ratios to determine the confidence set on the best model.
table: a reduced table with the models included in the confidence set.
Burnham, K. P., Anderson, D. R. (2002) Model Selection and Multimodel Inference: a practical information-theoretic approach. Second edition. Springer: New York.
Dail, D., Madsen, L. (2011) Models for estimating abundance from repeated counts of an open population. Biometrics 67 , 577--587.
MacKenzie, D. I., Nichols, J. D., Lachman, G. B., Droege, S., Royle, J. A., Langtimm, C. A. (2002) Estimating site occupancy rates when detection probabilities are less than one. Ecology 83 , 2248--2255.
MacKenzie, D. I., Nichols, J. D., Hines, J. E., Knutson, M. G., Franklin, A. B. (2003) Estimating site occupancy, colonization, and local extinction when a species is detected imperfectly. Ecology
84 , 2200--2207.
Royle, J. A. (2004) N-mixture models for estimating population size from spatially replicated counts. Biometrics 60 , 108--115.
Marc J. Mazerolle
AICc, aictab, c_hat, evidence, importance, modavg, modavgShrink, modavgPred
##anuran larvae example from Mazerolle (2006) data(min.trap) ##assign "UPLAND" as the reference level as in Mazerolle (2006) min.trap$Type <- relevel(min.trap$Type, ref = "UPLAND") ##set up candidate models Cand.mod <- list() ##global model Cand.mod[[1]] <- glm(Num_anura ~ Type + log.Perimeter + Num_ranatra, family = poisson, offset = log(Effort), data = min.trap) Cand.mod[[2]] <- glm(Num_anura ~ Type + log.Perimeter, family = poisson, offset = log(Effort), data = min.trap) Cand.mod[[3]] <- glm(Num_anura ~ Type + Num_ranatra, family = poisson, offset = log(Effort), data = min.trap) Cand.mod[[4]] <- glm(Num_anura ~ Type, family = poisson, offset = log(Effort), data = min.trap) Cand.mod[[5]] <- glm(Num_anura ~ log.Perimeter + Num_ranatra, family = poisson, offset = log(Effort), data = min.trap) Cand.mod[[6]] <- glm(Num_anura ~ log.Perimeter, family = poisson, offset = log(Effort), data = min.trap) Cand.mod[[7]] <- glm(Num_anura ~ Num_ranatra, family = poisson, offset = log(Effort), data = min.trap) Cand.mod[[8]] <- glm(Num_anura ~ 1, family = poisson, offset = log(Effort), data = min.trap) ##check c-hat for global model c_hat(Cand.mod[[1]]) #uses Pearson's chi-square/df ##note the very low overdispersion: in this case, the analysis could be ##conducted without correcting for c-hat as its value is reasonably close ##to 1 ##assign names to each model Modnames <- c("type + logperim + invertpred", "type + logperim", "type + invertpred", "type", "logperim + invertpred", "logperim", "invertpred", "intercept only") ##compute confidence set based on 'raw' method confset(cand.set = Cand.mod, modnames = Modnames, second.ord = TRUE, method = "raw") ##example with linear mixed model ## Not run: require(nlme) ##set up candidate model list for Orthodont data set shown in Pinheiro ##and Bates (2000: Mixed-effect models in S and S-PLUS. Springer Verlag: ##New York.) Cand.models <- list() Cand.models[[1]] <- lme(distance ~ age, random = ~age | Subject, data = Orthodont, method = "ML") Cand.models[[2]] <- lme(distance ~ age + Sex, data = Orthodont, random = ~ 1 | Subject, method = "ML") Cand.models[[3]] <- lme(distance ~ 1, data = Orthodont, random = ~ 1 | Subject, method = "ML") ##create a vector of model names Modnames <- paste("mod", 1:length(Cand.models), sep = "") ##compute confidence set based on 'raw' method confset(cand.set = Cand.models, modnames = Modnames, second.ord = TRUE, method = "raw") ##round to 4 digits after decimal point print(confset(cand.set = Cand.models, modnames = Modnames, second.ord = TRUE, method = "raw"), digits = 4) confset(cand.set = Cand.models, modnames = Modnames, second.ord = TRUE, level = 0.9, method = "raw") ##compute confidence set based on 'ordinal' method confset(cand.set = Cand.models, modnames = Modnames, second.ord = TRUE, method = "ordinal") ##compute confidence set based on 'ratio' method confset(cand.set = Cand.models, modnames = Modnames, second.ord = TRUE, method = "ratio", delta = 4) confset(cand.set = Cand.models, modnames = Modnames, second.ord = TRUE, method = "ratio", delta = 8) detach(package:nlme) ## End(Not run)
Useful links