binomial_double_trees

  • Available in: DRF

  • Hyperparameter: no

Description

When building classification models, this option specifies to build twice as many internal trees as the number of trees (one per class). Enabling this option can lead to higher accuracy but lower speed times, while disabling this can result in faster model building. This option is disabled by default.

Note that ntrees=50 by default, so specifying binomial_double_trees=TRUE without specifying a number of trees will result in 50 trees and 100 internal trees.

Example

R

Python

library(h2o)
h2o.init()

# import the cars dataset:
# this dataset is used to classify whether or not a car is economical based on
# the car's displacement, power, weight, and acceleration, and the year it was made
cars <- h2o.importFile("https://s3.amazonaws.com/h2o-public-test-data/smalldata/junit/cars_20mpg.csv")

# convert response column to a factor
cars["economy_20mpg"] <- as.factor(cars["economy_20mpg"])

# set the predictor names and the response column name
predictors <- c("displacement","power","weight","acceleration","year")
response <- "economy_20mpg"

# split into train and validation sets
cars_splits <- h2o.splitFrame(data =  cars, ratios = 0.8, seed = 1234)
train <- cars_splits[[1]]
valid <- cars_splits[[2]]

# try using the `binomial_double_trees` (boolean parameter):
car_drf <- h2o.randomForest(x = predictors, y = response, training_frame = train,
                   validation_frame = valid,
                   binomial_double_trees = FALSE, seed = 1234)

# print the auc and the number of trees built with binomial_double_trees turned off
# see the difference between the number of trees and number of internal trees
print(paste('without binomial_double_trees', h2o.auc(car_drf, valid = TRUE), sep = ": "))
print(car_drf@model$model_summary$number_of_trees)
print(car_drf@model$model_summary$number_of_internal_trees)


# try using the `binomial_double_trees` (boolean parameter):
car_drf_2 <- h2o.randomForest(x = predictors, y = response, training_frame = train,
                   validation_frame = valid,
                   binomial_double_trees = TRUE, seed = 1234)

# print the auc and the number of trees built with binomial_double_trees turned on
# see the difference between the number of trees and number of internal trees
print(paste('with binomial_double_trees', h2o.auc(car_drf_2, valid = TRUE), sep = ": "))
print(car_drf_2@model$model_summary$number_of_trees)
print(car_drf_2@model$model_summary$number_of_internal_trees)

import h2o
from h2o.estimators.random_forest import H2ORandomForestEstimator
h2o.init()


# import the cars dataset:
# this dataset is used to classify whether or not a car is economical based on
# the car's displacement, power, weight, and acceleration, and the year it was made
cars = h2o.import_file("https://s3.amazonaws.com/h2o-public-test-data/smalldata/junit/cars_20mpg.csv")

# convert response column to a factor
cars["economy_20mpg"] = cars["economy_20mpg"].asfactor()

# set the predictor names and the response column name
predictors = ["displacement","power","weight","acceleration","year"]
response = "economy_20mpg"

# split into train and validation sets
train, valid = cars.split_frame(ratios = [.8], seed = 1234)

# try using the binomial_double_trees (boolean parameter):
# Initialize and train a DRF
cars_drf = H2ORandomForestEstimator(binomial_double_trees = False, seed = 1234)
cars_drf.train(x = predictors, y = response, training_frame = train, validation_frame = valid)

# print the auc and the number of trees built with binomial_double_trees turned off
print('without binomial_double_trees:', cars_drf.auc(valid=True))


# Initialize and train a DRF
cars_drf_2 = H2ORandomForestEstimator(binomial_double_trees = True, seed = 1234)
cars_drf_2.train(x = predictors, y = response, training_frame = train, validation_frame = valid)

# print the auc and the number of trees built with binomial_double_trees turned on
print('with binomial_double_trees:', cars_drf_2.auc(valid=True))