Data Science with R: Decision Trees and Random Forests/

...

/

Tuning a Classification Tree

Tuning a Classification Tree

Build on cross-validation by learning to tune a CART classification decision tree using tidymodels.

We'll cover the following...

Preparing the data

The following code prepares the Titanic training data as part of a tidymodels workflow.

R
#================================================================================================
# Load libraries - suppress messages
#
suppressMessages(library(tidyverse))
suppressMessages(library(tidymodels))
suppressMessages(library(rattle))
#================================================================================================
# Load the Titanic training data and transform Embarked to a factor
#
titanic_train <- read_csv("titanic_train.csv", show_col_types = FALSE) %>%
  mutate(Sex = factor(Sex),
         Embarked = factor(case_when(
           Embarked == "C" ~ "Cherbourg",
           Embarked == "Q" ~ "Queenstown",
           Embarked == "S" ~ "Southampton",
           is.na(Embarked) ~ "missing")))
#================================================================================================
# Craft the recipe - recipes package
#
titanic_recipe <- recipe(Survived ~ Sex + Pclass + SibSp + Parch + Fare + Embarked, data = titanic_train) %>%
  step_num2factor(Survived,
                  transform = function(x) x + 1,
                  levels = c("perished", "survived")) %>%
  step_num2factor(Pclass,
                  levels = c("first", "second", "third"))

Configuring the model for tuning

The parsnip package supports a variety of methods for tuning hyperparametersA parameter whose value is used to control the learning process.. Configuration of a tidymodels workflow for hyperparameter tuning occurs when the algorithm is specified. The parsnip package’s decision_tree() function supports tuning of the following hyperparameters:

  • cost_complexity: A positive number for the cost/complexity parameter (aka cp) used by the rpart package.

  • tree_depth: A positive integer for the maximum depth of the tree.

  • min_n: A positive integer for the minimum number of data points in a node required for the node to be split further. This corresponds to ...