A decision tree is a non-parametric supervised learning algorithm which can be used for both classification and regression. Decision trees are like inverted trees with the root node at the top and leaf nodes forming downwards. There are different algorithms to split the dataset into branch-like segments. Each leaf node is assigned to a class that represents the most appropriate target values.

Decision trees do not require any scaling or transformations of the dataset and work as the data is. They can handle both categorical and continuous features, and also address non-linearity in the dataset. At its core, a decision tree is a greedy algorithm (it considers the best split at the moment and does not take into consideration the future scenarios) that performs a recursive binary partitioning of the feature space. Splitting is done based on information gain at each node because information gain measures how well a given attribute separates the training examples as per the target...