Session

Computer Science and Communication Engineering

Description

Classification and regression tree (CART) is a non-parametric methodology that was introduced first by Breiman and colleagues in 1984. CART is a technique which divides populations into meaningful subgroups that allows the identification of groups of interest. CART as a classification method constructs decision trees. Depending on information that is available about the dataset, a classification tree or a regression tree can be constructed. The first part of this paper describes the fundamental principles of tree construction, pruning procedure and different splitting algorithms. The second part of the paper answers the questions why or why not the CART method should be used or not. The advantages and weaknesses of the CART method are discussed and tested in detail. Finally, CART is applied to an example with real data, using the statistical software R. In this paper some graphical and plotting tools are presented.

Keywords:

Classification, Regression, Tree, Pruning, split, Goodness of fit, Algorithm

Proceedings Editor

Edmond Hajrizi

ISBN

978-9951-437-46-2

First Page

38

Last Page

54

Location

Durres, Albania

Start Date

28-10-2016 9:00 AM

End Date

30-10-2016 5:00 PM

DOI

10.33107/ubt-ic.2016.52

Download

Share

COinS
 
Oct 28th, 9:00 AM Oct 30th, 5:00 PM

A SUMMARY OF Classification and Regression Tree WITH APPLICATION

Durres, Albania

Classification and regression tree (CART) is a non-parametric methodology that was introduced first by Breiman and colleagues in 1984. CART is a technique which divides populations into meaningful subgroups that allows the identification of groups of interest. CART as a classification method constructs decision trees. Depending on information that is available about the dataset, a classification tree or a regression tree can be constructed. The first part of this paper describes the fundamental principles of tree construction, pruning procedure and different splitting algorithms. The second part of the paper answers the questions why or why not the CART method should be used or not. The advantages and weaknesses of the CART method are discussed and tested in detail. Finally, CART is applied to an example with real data, using the statistical software R. In this paper some graphical and plotting tools are presented.