Decision Trees example

We examine the cube Sales from the database Foodmart2000 and want to build a Decision Tree, which can predict the yearly income of the customers in accordance with their sex, marital status, education level and member card. In Decision Trees module of BI2M application you have to follow the steps:

Step 1 - In the given cube we have Country, State Province, City and Name as levels of the dimension Customers. As we will analyze different clients, we choose Name.

Step 2 – choose the level to be predicted - in this case Yearly Income.

Step 3 – choose the data which the Decision Trees algorithm will process to find results – in the current case they are Gender, Marital Status, Education Level and Member Card.

The result is:

The results of the example show that the most important characteristic for Yearly Income prediction is Education and then Member Card. A branch of Prediction Tree represents all the cases of customers with a definite kind of education and/or member card. On choosing of a branch the numerical values are represented in a table. The table in the current example represents kinds of yearly income in the database. The table shows the number of customers of the examined branch with corresponding yearly income and the probability a customer of the branch to be with such yearly income. The results in the table are illustrated by a histogram.