Explore a wider model space

Machine Learning with caret in R

Zach Mayer

Data Scientist at DataRobot and co-author of caret

Random forests require tuning

Hyperparameters control how the model is fit
Selected "by hand" before the model is fit
Most important is mtry
- Number of randomly selected variables used at each split
Lower value = more random
Higher value = less random
Hard to know the best value in advance

Example: sonar data

tuneLength argument to caret::train()
Tells caret how many different variations to try

# Load some data
library(caret)
library(mlbench)
data(Sonar)

# Fit a model with a deeper tuning grid
model <- train(
  Class ~ ., 
  data = Sonar, 
  method = "ranger", 
  tuneLength = 10
)

# Plot the results
plot(model)

Plot the results

The image contains a line plot of bootstrap accuracy versus the number of randomly selected parameters. It rises to a peak around x = 14, then drops off quickly.

Let's practice!

Machine Learning with caret in R