Gini Impurity vs Information Gain vs Chi-Square - Methods for Decision Tree Split

Decision trees are one of the most used machine learning models because of their ease of implementation and simple interpretations. To better learn from the data they are applied to, the nodes of the decision trees need to be split based on the attributes of the data. In this article, we will understand the need of splitting a decision tree along with the methods used to split the tree nodes. Gini impurity, information gain and chi-square are the three most used methods for splitting the decision trees. Here we will discuss these three methods and will try to find out their importance in specific cases.