Advances and Applications in Statistics
Volume 45, Issue 3, Pages 201 - 223
(June 2015) http://dx.doi.org/10.17654/ADASJun2015_201_223 |
|
CONSTRUCTIVE REGRESSION ON IMPLICIT REGIONS
Jessica Kubrusly and Hélio Lopes
|
Abstract: Tree-based methods are playing an important role in non-linear regression. Starting from the CART algorithm introduced in the 80’s, several variants have been proposed, aiming at improving existing models. This work introduces a new tree-based method called constructive regression on implicit regions, where the input domain space is partitioned differently from the other tree-based methods. The proposed algorithm allows partitions formed by flexible regions and returns a continuous regression due to the use of transition functions. This new method shows competitive results when compared to other regression methods and applied to well data. |
Keywords and phrases: regression trees, nearest neighbors, algorithms. |
|
Number of Downloads: 440 | Number of Views: 1331 |
|