H2O AutoML Leaderboard Utilities

The H2O AutoML Leaderboard lists any models that have been created during the automl_reg() training process.

• The training process automatically uses the top model.

• The available models can be shown with automl_leaderboard()

• The model change the model used using automl_update_model().

Usage

automl_leaderboard(object)

automl_update_model(object, model_id)

Arguments

object

An object created by automl_reg() and trained (fitted).

model_id

An H2O Model ID (shown in the AutoML Leaderboard). Alternatively, the user can provide an H2O model.

Value

• automl_leaderboard(): A tibble containing the H2O AutoML Leaderboard

• automl_update_model(): An updated parnsip or workflow with the H2O Model updated

Examples

if (FALSE) {
library(tidymodels)
library(modeltime.h2o)
library(h2o)
library(tidyverse)
library(timetk)

h2o.init(
    nthreads = -1,
    ip       = 'localhost',
    port     = 54321
)

# Model Spec
model_spec <- automl_reg(mode = 'regression') %>%
    set_engine(
        engine                     = 'h2o',
        max_runtime_secs           = 5, 
        max_runtime_secs_per_model = 4,
        nfolds                     = 5,
        max_models                 = 3,
        exclude_algos              = c("DeepLearning"),
        seed                       = 786
    ) 


# Fit AutoML
model_fit <- model_spec %>%
    fit(value ~ ., data = training(m750_splits))

# Inspect the Leaderboard
leaderboard_tbl <- automl_leaderboard(model_fit)
leaderboard_tbl

# Swap an H2O Model Out (Using the 2nd model from the leaderboard)
model_id_2  <- leaderboard_tbl$model_id[[2]]
model_fit_2 <- automl_update_model(model_fit, model_id_2)
model_fit_2

# Shutdown H2O when Finished. 
# Make sure to save any work before. 
h2o.shutdown(prompt = FALSE)
    
    
}