get info of significant variables in each node of a ctree

see @return for details. Note that package partykit is loaded, but not attached, in this function.

Usage

ctree2splitInfo(cRe, pCut, nodeInds = partykit::nodeids(cRe, terminal = F))

cRe: object returned from partykit::ctree()
pCut: numeric; p-val cutoff to be significant
nodeInds: int vectors, e.g. 1:5 for a tree of five nodes, including leaves

return a list of following items:

pVals is a list of significant variables at each node of nodeInds excluding leaves;
- length(pVals): number of internal nodes
- names(pVals): the name of splitting variable at each node
- each item of pVals is a numeric vector of p-vals for all variables whose p-val <= pCut, ordered by increasing p-vals
spVars.df a data.frame of following four columns, one node per row,
- variable: the name of splitting variable (spVar)
- pVal: the p-val of the spVar
- sCut: the splitting point
- gtOnRight: T/F; are observations of greater (gt: greater than) values on the right branch of the node? The separation of numeric spVar is always <= and >; if the leaves are listed in increasing y, gtOnRight==T means the spVar has positive association with y.
spVars.str, a string combining all rows of spVars.df. rows are separated by ';' and cells are by ','

if cRe has only one root node, i.e. no splitting, NULL is returned

# none