pirouette: pirouette

In zachmayer/pirouette: Fast Rotation Forests

Description

Random rotation forests for sparse data

Usage

pirouette(x, y, prob = 1/sqrt(ncol(x)), ctrl = pirouetteControl(),
  gbm_control = list(n.trees = 1, interaction.depth = 2, shrinkage = 0.1,
  verbose = FALSE), ...)

Arguments

x	a sparse matrix of x variables
y	the target variable for classification or regression
prob	the probability each entry of the rotation matrix is non-zero. 1/3 is one heuristic to use (for 3x faster processing), 1/sqrt(ncol(x)) is another, for sqrt(x)-fold faster processing. Passed through enpointe to splat.
ctrl	a list of control parameters for the algorithm
gbm_control	a list of control parameters for the gbm
...	passed through enpointe to gbm

Value

an object of class pirouette

References

• https://stat.ethz.ch/pipermail/r-sig-hpc/2013-January/001575.html

• http://web.stanford.edu/~hastie/Papers/Ping/KDD06_rp.pdf

• http://stats.lse.ac.uk/fryzlewicz/rre/rre.pdf

zachmayer/pirouette documentation built on May 4, 2019, 9:04 p.m.