conclude.model: Draw Conclusions on Models from a "CAM" class object
In QIU-Hongxiang-David/CAMer: Continuous Admixture Modeler

Description Usage Arguments Details Value Note See Also Examples

View source: R/CAMer.R

Draw conclusions on which models are best from a "CAM" class object or its summary table.

1	conclude.model(x, alpha = 0.05, p.adjust.method = "holm", log = TRUE)

`x`	a "CAM" class object or the summary table of a "CAM" class object
`alpha`	familywise type-I error rate. Defaults to 0.05
`p.adjust.method`	method for adjusting p-values to adapt for familywise type-I error rate. Defaults to `"holm"`
`log`	a logical expression. Whether log transformation should be applied to msE. Defaults to `TRUE` to make the null distribution more symmetric.

The function uses pairwise Wilcoxon signed-rank test on pseudo log(msE) or msE based on Jackknives to select the best model(s). These pseudovalues are treated as if they were independent following Tukey (1958).

If HI model is not significantly worse than any other model, it is chosen as the best model; otherwise, the model(s) with significantly smallest msE are chosen as best model(s).

The estimated interval is the one that include all time points covered by more than half of the intervals estimated from Jackknives. The estimated point (fot HI model) is the nearest integer to the mean of the points estimated from Jackknives.

There is a special print method for this class.

an object of S3 class "CAM.conclusion". A list consisting of:

`call`	function call
`group.means`	a named vector of group means of pseudo log(msE)/msE with each model being a group
`adjusted.p.value`	a matrix of adjusted p-values for pairwise differences with the i,j-th entry being the adjusted p-value for the difference in pseudo log(msE)/msE of Model i and Model j. All entries on the diagonal are `NA`.
`best.models`	a data frame consisting of best models concluded and their estimated time intervals/points
`pseudovalue`	a data frame containing all pseudovalues, i.e. pseudo msE/log(msE)
`p.adjust.method`	method for adjusting p values used

The function only chooses the best model(s). It does NOT do any diagnostic analysis. Particularly, it does NOT check whether the best model(s) are credible.

CAM, p.adjust, pairwise.t.test

data(CGF_50)
fit<-CAM(CGF_50,0.3,70L,isolation=FALSE)
conslusion<-conclude.model(fit)

conslusion<-conclude.model(fit,alpha=0.01,p.adjust.method="bonferroni",log=FALSE)