Usage
lrtree.control(minsplit = 20, minbucket = round(minsplit/2), maxdepth = 5, eps = 3.84, WH.strd = 1,
fitted.model = c("logrank","cart"),
tree.size = c("Boot", "ML"), B = 200,
random = FALSE, plot = FALSE,
rho = 0, pre.pt = NA,
m.zero = 3, Jt = 2, classwt = FALSE, L.split = FALSE, time.varying = FALSE, ncl = 2
)Arguments
minsplit
the minimum number of subjects that must exist in a node, in order for
a split to be attempted.
minbucket
the minimum number of observations in any terminal node.
If only one of minbucket or minsplit is specified,
the code either sets minsplit to minbucket*3
or minbucket to
maxdepth
the maximum of tree depth. Default is 5.
eps
a numerical criteria in split variable statistic in order to stop growing tree.
WH.strd
a numerical criteria in WH approximation in order to stop growing tree .
fitted.model
a parameter for a splitting and growing altorithm:
"logrank" - fit Segal's log rank survival tree
"cart" - fit survival cart algrotihm.
tree.size
a parameter to determine the size of the tree:
"Boot" - use a bootstrapping split rule at each split.
"ML" - use Multi-Level direct stopping rule (Eo and Cho, 2012+).
"BU" - use the Bottom-Up approach (Segal, 1988).
B
a scalar parameter to choose the number of bootrapping at each split.
random
a logical parameter to treat multiple responses censored data.
plot
a logical parameter to draw test statistic at split node t.
rho
a numeric parameter to choose the type of methos in survfit function.
pre.pt
predetermined split point for split point selection.
m.zero
a parameter to "RG" methods. Only splits yielding at least m.zero cases in each subnode.
Jt
numerical values to set criterion for constructing variable selection table.
classwt
Priors of the classes. Need not add up to one. Ignored for regression.
L.split
Logical parameter to use Linear splits. If TRUE, run linear split test.
time.varying
Logical parameter to treat time varying predictors.
ncl
the number of cores to use for cost-complexity pruning.