visualiseGroupweights: Visualise estimated group weights

In ecpc: Flexible Co-Data Learning for High-Dimensional Prediction

Visualise estimated group weights

Description

Plot group weights from multiple cross-validation folds.

Usage

visualiseGroupweights(dfGrps, Groupset, groupset.grouplvl, values, 
                      widthBoxplot = 0.05, boxplot = TRUE, jitter = TRUE, 
                      ps = 1.5, ls = 1)

Arguments

dfGrps	Data frame containing the following variables; 'Group': factor with group names; 'Group.weight': group weight of each group; 'Fold': number indicating which fold in the cross-validation is used.
Groupset	List of G elements containing covariate indices for each group
groupset.grouplvl	(optional): groups on group level, e.g. defining a hierarchical structure.
values	(optional): values of continuous co-data. If given, group weights are plotted against these value.
widthBoxplot	Width of boxplot.
boxplot	Should a boxplot be plotted?
jitter	Should point estimates be jittered?
ps	Point size.
ls	Line size.

Value

Plot in ggplot object.

See Also

visualiseGroupset to visualise group sets and visualiseGroupsetweights to plot estimated group set weights.

Examples

#discrete groups
groupset1 <- list(1:20,21:40)
dfGrps1 <- data.frame(Group=as.factor(rep(c(1,2),each=10)),
                     Group.weight=c(rnorm(10,0.5,0.01),rnorm(10,2,0.05)),
                     Fold=rep(1:10,2))
visualiseGroupweights(dfGrps1, Groupset=groupset1)

#continous co-data groups
cont.codata <- seq(0,1,length.out=40) #continuous co-data
#only split at lower continous co-data group
groupset2 <- splitMedian(values=cont.codata,split="lower",minGroupSize=10)
#obtain groups on group level defining the hierarchy
groupset.grouplvl <- obtainHierarchy(groupset2)

#simulate random group weights around 1
dfGrps2 <- data.frame(Group=as.factor(rep(1:length(groupset2),each=10)),
                      Group.weight=c(rnorm(10*length(groupset2),1,0.01)),
                      Fold=rep(1:10,length(groupset2)))
#plot group weights per group
visualiseGroupweights(dfGrps2, Groupset=groupset2, groupset.grouplvl=groupset.grouplvl)
#plot group weights per leaf group in the hierarchical tree
visualiseGroupweights(dfGrps2, Groupset=groupset2, groupset.grouplvl=groupset.grouplvl,
                      values=cont.codata)

ecpc documentation built on March 7, 2023, 6:46 p.m.