Mastering Machine Learning with Spark 2.x

Trees

With the definition of connected components in our hands, we can turn to another interesting class of graphs, namely trees. A tree is a connected graph in which there is precisely one path connecting any given vertex to another. A graph consisting of a disjointed group of trees is called a forest. In the following diagram, we see a schematic decision tree ran on the well known Iris dataset. Note that this is for illustration purposes only, that is, to show how the output of this algorithm can be seen as a graph:

Figure 7: A simple decision tree ran on Iris, classifying into the three categories Setosa, Virginica and Versicolor by means ...