Decision trees

In this vid, we’ll manually construct a decision tree classifier using the following data set.

x	y	color
7.8	6.1	blue
13.7	4.5	red
21.0	5.7	blue
24.6	14.8	blue
18.9	21.8	red
5.0	14.8	red
13.4	17.1	red
32.3	2.8	blue
34.9	20.3	blue
32.6	12.3	red

Decision trees

A decision tree classifier applies a divide-and-conquer algorithm to the training set, recursively splitting it into pieces that are as homogeneous as possible.

Here’s how I made that plot:

ggplot(dt_example, aes(x, y, color = color)) +
  geom_point(size = 7,
             show.legend = FALSE) +
  labs(x = NULL,
       y = NULL) + 
  theme(axis.text = element_text(size = 15,
                                 face = "bold")) +
  scale_color_manual(values = c("blue", "red"))