DECISION TREE EXAMPLES

Decision trees nodes and structure vary depending on the object of data mining and on the structure of information you possess.

Below  there are three decision tree examples that will help you to understand the idea of decision tree technique and its application in data mining.

DECISION TREE Example #1 -  beginner

To understand the idea of decision tree lets start with a simple example: object exploration.

Lets suppose we have some information about three objects: a box, a ball, and a clew. The goal is to create a decision tree using which we will be able to define: whether the object is a box, a ball, or a clew? Such a process is called classification.

The initial information is:

• objects form - is it round (Yes/No)
• structure - does it contain threads (Yes/No).

Lets create a decision tree that will determine what an object is. In case if we select "objects form" as a root node, we will receive:

In other case the decision tree will start with "structure":

In both examples we can see a path from the root node up to the final node. Now, the created tree can be used for classifying objects: moving through the tree nodes, we will be able to determine objects type. For example: if (the object is round= Yes) then if (does it contain threads = No) then (Object Class = Ball).

The example might seem too easy, but it is important to understand, that just as classifying such simple objects, you will be able to classify your clients, business processes, marketing campaigns and many other vital information - a power to forecast your business future.

DECISION TREE Example #2 - marketing decision tree

One of well-know decision trees application is marketing.

The goal is to create an extremely effective targeted marketing campaign with maximized return on investment (ROI).

DECISION TREE Example #3 - insurance decision tree

Decision trees method can be used not only in marketing, but in any other field that involves decision making process. And if in case of marketing campaign, we will need a decision tree from time to time, this doesn't mean that this data mining method cannot be of everyday use.

A good example of possible everyday use are decision trees in insurance industry. In this case, the key question is: "does the client belong to the reliable, not fraudulent group?"

The initial information can be :

• sex (M/F)
• age (under 25, 26-35, 36-45, 46-55, over 56)
• education ()
• marital status (Yes(married)/No(not married))
• real estate (Yes(is owner)/No(doesn't own))
• approximate annual revenue
• belonging to fraudulent group

A part of the decision tree could look like this:

This kind of decision tree allows to quickly analyze to what group a client might belong. Such analysis of clients database will help with making the decision about insuring. Now the decision can be based not only on intuition, but on reliable facts.

Please remember, that these are only examples of decision trees. These examples can not be used for analysis of your databases. To receive your own decision trees use Estard Data Miner.