www.allitebooks.com

Chapter 3
You can change those indices to look at other parts of the data, as there are over 1000
games in our dataset!
Currently, this gives a false value to all teams (including the previous year's
champion!) when they are first seen. We could improve this feature using the
previous year's data, but will not do that in this chapter.
Decision trees
Decision trees are a class of supervised learning algorithm like a flow chart that
consists of a sequence of nodes, where the values for a sample are used to make a
decision on the next node to go to.
As with most classification algorithms, there are two components:
• The first is the training stage, where a tree is built using training data.
While the nearest neighbor algorithm from the previous chapter did not
have a training phase, it is needed for decision trees. In this way, the nearest
neighbor algorithm is a lazy learner, only doing any work when it needs
to make a prediction. In contrast, decision trees, like most classification
methods, are eager learners, undertaking work at the training stage.
• The second is the predicting stage, where the trained tree is used to
predict the classification of new samples. Using the previous example
tree, a data point of ["is raining", "very windy"] would be classed
as "bad weather".
[ 47 ]
www.allitebooks.com