Page 64 - Data Science Algorithms in a Week

Decision Trees

Swim preference - representing data with

decision tree

For example, we may have certain preferences on whether we would swim or not. This can
be recorded in the table as follows:

Swimming suit Water temperature Swim preference
None Cold No

None Warm No
Small Cold No
Small Warm No

Good Cold No
Good Warm Yes

Data in this table can be represented alternatively with the following decision tree, for
example:

Figure 3.1.: Decision tree for the swim preference example

At the root node, we ask the question: does one have a swimming suit? The response to the
question separates the available data into three groups, each with two rows. If the attribute
swimming suit = none, then two rows have the attribute swim preference as no.
Therefore, there is no need to ask a question about the temperature of the water as all the
samples with the attribute swimming suit = none would be classified as no. This is also
true for the attribute swimming suit = small. In the case of swimming suit = good,
the remaining two rows can be divided into two classes: no and yes.

[ 52 ]