gbm_grid: CV Tune gbm fits over a grid of parameters in parallel

In patr1ckm/mvtboost: Tree Boosting for Multivariate Outcomes

Description

This function cross validates a grid of meta-parameter arguments implied by expand.grid(...). The final model is returned with the best fitting parameters.

Usage

gbm_grid(y, x, cv.folds, mc.cores = 1, subset = NULL, ...)

Arguments

y	vector of outcomes
x	matrix of predictors
cv.folds	number of cross-validation folds
mc.cores	number of cores
subset	subset of the data to use for training; cross-validation will be performed within subset
...	args to gbm.fit. Arguments can be passed vectors. The function tunes across all rows of expand.grid(...)

Details

Non-grid arguments (such as distribution, weights, etc) can be passed as normal.

Value

A list with $gbm (fitted model with best meta parameters), $best_args (best meta parameters), and $args (all meta parameters with cv err)

patr1ckm/mvtboost documentation built on May 24, 2019, 8:21 p.m.