Recursive Partitioning for Supervised Learning

first distribution means

x

4.6

y

0.8

z

9.7

second distribution means

x

9.4

y

4.9

z

4.2

number of samples

n

30

Data is sampled from two three-dimensional Gaussian distributions with means that you can vary. Data points from the first Gaussian are labeled

A

and data points from the second Gaussian are labeled

B

. The recursive partitioning algorithm is run on this training data to determine rules for classifying new data points. The first plot shows the training data and the regions that are determined by the algorithm. The bar chart at the top right gives a visual representation of the means of the Gaussian distributions. The tree at the bottom is a visual representation of the classification rules determined by the algorithm. Each path to a leaf node represents a region in space.