Classification is a process that is broadly divided into two steps. This is the learning step and the step where prediction is done. In the stage of learning the model gets developed on the training data that is given. In the prediction stage, the model is used in order to predict the response to the data given. A very popular and easy way of the classification trees algorithm is the decision tree which is not just easy to understand but also easy to interpret.

  1. What is CT?
  2. How do classification trees work?
  3. CT Example

1) What is CT?  

The decision tree method is recommended when the task of data mining contains the prediction or classification of the outcome. The goal is to generate rules that can be explained easily. This can then be translated to the natural query language or to SQL.

The classification trees will label, record, and assign the variable to the discrete classes. The classification tree will also provide a confidence measure that the classification is in fact correct. The classification tree gets built using a process of binary recursive partitioning. This process is iterative by splitting the data into various partitions. It is then split up further on each of the branches.

2) How do classification trees work?

Here you need to understand the working of the classification tree. You first need to make predictions with the classification trees. It is best to predict the numerical target or the category with the classification tree. This is one of the main benefits of using this algorithm. All that is needed is to start at the root node. You then need to look at the feature value and then evaluate. Depending on the value you go to either the left or to the right of the children node.

The process gets repeated until you reach the leave node. When this occurs then depending on the classification problem the predicted category will be the mode of the categories on the leave node. A test sample that will reach this node has the maximum probability of belonging to the class with the training samples.

For the regression tree, the prediction is made at the end and this is the mean of the values for the variable that thus targets at the leave node.

In the first step, the training set gets created where the classification label is known for every record. The algorithm will systematically assign each record to one of the two subsets that are available on the basis of some factor. This helps to get a set of homogenous labels in every partition. The splitting is applied to every new partition and the process will continue till there are no more splits found.

3) CT Example 

The algorithm of the classification decision trees will belong to the family of the supervised learning algorithm. Unlike the other learning algorithms that are supervised the decision tree algorithm can be used to solve the classification and the regression problems. The main aim to use the decision tree is to create a training model that can be used to predict the value or class of the variable that is targeted.

This is done by learning the simple decision rules that are inferred from any prior data. In the decision tree to predict the class label for the record one starts with the root of a tree. The values of the root are compared to the attribute of the records. On the basis of this comparison, the branch is followed corresponding to the value, and then there is a jump to the next node.

A classification tree example could be when the company has to predict if the customer will pay his renewal premium. The income of the customer is significant to consider but again the insurance company may not have the salary details of each customer. Since salary is a crucial variable a classification tree can be built to predict the customers’ income based on his occupation and other variables. This helps to predict the value of a variable that is continuous in nature.


The classification tree is an algorithm that is intuitive and simple. Because of this, it is used a lot to try and explain the machine learning model results. The classification tree can be used to make boosting and bagging models which in turn is very powerful.

If you are interested in making it big in the world of data and evolve as a Future Leader, you may consider our Integrated Program in Business Analytics, a 10-month online program, in collaboration with IIM Indore!



Are you ready to build your own career?