LinCDE.boost.helper: LinCDE.boost.helper

In ZijunGao/LinCDE: Conditional Density Estimation via Lindsey's Method and Boosting

LinCDE.boost.helper

Description

This function is a helper for the function LinCDE.boost. This helper grows one LinCDE tree with heterogeneous carrier density.

Usage

LinCDE.boost.helper(
  X,
  yIndex,
  z = NULL,
  currDepth,
  tree,
  indexList,
  currNodeIndex,
  cellProb,
  improvementQueue,
  splitPoint,
  numberBin = 20,
  splineDf = 10,
  lambda = 0,
  penalty = NULL,
  df = 6,
  alpha = 0.2,
  terminalSize = 20,
  depth = 1,
  shrinkage = 0.1
)

Arguments

X	input matrix, of dimension nobs x nvars; each row represents an observation vector.
yIndex	discretized response vector, of length nobs.
z	sufficient statistics matrix, of dimension numberBin x splineDf. Default is NULL.
currDepth	the current depth of the LinCDE tree.
tree	the current LinCDE tree.
indexList	the list of candidate splits.
currNodeIndex	the index of the current node.
cellProb	cell probability matrix, of dimension nobs x numberBin.
improvementQueue	the queue recording improvements of candidate splits.
splitPoint	a list of candidate splits, of length nvars.
numberBin	the number of bins for response discretization. Default is 20.
splineDf	the number of sufficient statistics. Default is 10.
lambda	the regularization parameter for Poisson regression. Default is 0.
penalty	separate penalty factors applied to each coefficient, of length splineDf. Default is NULL.
df	degrees of freedom. Default is 6.
alpha	parameter for the stopping rule of splitting. Default is 0.2.
terminalSize	the minimum number of observations in a terminal node. Default is 20.
depth	the number of splits of each LinCDE tree. The number of terminal nodes is depth + 1. For depth = 1, an additive model is fitted. Default is 1.
shrinkage	the shrinkage parameter applied to each tree in the expansion, value in (0,1]. Default is 0.1.

Value

This function returns a LinCDE tree trees and the updated cell probability matrix cellProb.

ZijunGao/LinCDE documentation built on Jan. 2, 2023, 11:14 p.m.