elbow.tree {TimeVTree}

Finding the Final Tree using the Elbow Method

Description

elbow.tree is like final.tree, but instead of using the minimum cost it uses the 'elbow' of the costs. It is similar to the elbow AIC or BIC approaches in the literature.

Usage

elbow.tree(nodetree=nodetree, subtrees=subtrees, omega, alphac=2)

Arguments

nodetree

Fully grown tree from the original data. Output from output.coxphout

subtrees

Pruned subtrees from the original data. Output from prune

omega

Bias (i.e. third index of the output) from bootstrap. Look at the value section of bootstrap for more information.

alphac

Predetermined penalty parameter

Details

One can take the output (table) generated by this function and plot the (penalized) bias-corrected cost of each subtrees, then (visually) identify the 'elbow' as the selected subtree.

Value

subtree

output from prune with an additional column 'cost' that contains bootstrap estimate of each subtree

cost.p

This column contains the (penalized) bias-corrected cost of each subtree

Examples

## Not run: 
data('alcohol')
require(survival)

coxtree <- coxph.tree(alcohol[,'time'], alcohol[,'event'], 
                      x = alcohol[,'alc', drop = FALSE], D = 4)
nodetree <- output.coxphout(coxtree)

subtrees <- prune(nodetree)

store.mult.cont <- bootstrap(B=20, nodetree, subtrees, alcohol[,'time'],
                                alcohol[,'event'], x = alcohol[,'alc', drop = FALSE], 
                                D=4,minfail=20, alphac=2)
                                
Balph <- 0.5 * 2 * log(nrow(alcohol))                                
elbow.tree <- elbow.tree(nodetree, subtrees, store.mult.cont[[3]], alphac= Balph)

## End(Not run)