Machine Learning for OpenCV

Understanding the decision rules

In order to build the perfect tree, you would want to split the tree at the most informative feature, resulting in the purest daughter nodes. However, this simple idea leads to some practical challenges:

• It's not actually clear what most informative means. We need a concrete value, a score function, or a mathematical equation that can describe how informative a feature is.
• In order to find the best split, we have to search over all the possibilities at every decision node.

Fortunately, the decision tree algorithm actually does these two steps for you. Two of the most commonly used criteria that scikit-learn supports are the following:

• criterion='gini': The Gini impurity is a measure of misclassification, ...