View source: R/h2otools_grid_CV.R
Optimizes 1 parameter. Algorithm tries to avoid over-fitting. So, it does not optimizes test results. Instead, it looks for overlapping of train and test metrics.
Check h2o.cv.rnd.opt.param.get_best
for select best parameter from output from this function.
Check h2o.ggplot.dfResultsCV
for ploting output results from this function.
1 2 3 4 5 6 7 8 9 | h2o.cv.rnd.opt.param(dataset.h2o, split.perc = 0.6,
model_algorithm = h2o.gbm, nfolds = 10, max_it = 10, opt.area = 0.5,
opt.tol = 0.01, opt.var = "shrinkage", opt.val = 0.1, opt.val.min = 0,
opt.val.max = 1, opt.val.is.integer = FALSE,
opt.criteria = "overfitting", opt.val.rang.perc = 0.9,
model_params = list(shrinkage = 0.1, n.trees = 30, interaction.depth = 5,
n.bins = 50, n.minobsinnode = 1000, distribution = "bernoulli", importance =
FALSE, balance.classes = TRUE), gain.calculate = TRUE, gain.ngroups = 100,
col.pred = 3, col.target = tail(h2o::colnames(dataset.h2o), n = 1))
|
h2o.cv.rnd.grid.shrinkage
h2o.cv.rnd.opt.param
h2o.cv.rnd.opt.param.get_best
h2o.ggplot.dfResultsCV
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.