|
Activity Number:
|
461
|
|
Type:
|
Invited
|
|
Date/Time:
|
Thursday, August 7, 2008 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
IMS
|
| Abstract - #300148 |
|
Title:
|
Regression Models for Symbolic Interval-Valued Data
|
|
Author(s):
|
Francisco de A.T. De Carvalho*+ and Eufrasio de A. Lima Neto
|
|
Companies:
|
Centro de Informatica - CIn/UFPE and Centro de Informatica - CIn/UFPE
|
|
Address:
|
Av. Prof. Luiz Freire, s/n - Cidade Universitaria, Recife-PE, 50740540, Brazil
|
|
Keywords:
|
Symbolic Data Analysis ; Symbolic Interval-Valued Data ; Regression
|
|
Abstract:
|
This presentation concerns regression models for symbolic interval-valued data. In the framework of Symbolic Data Analysis (Billard and Diday (2007)), Billard and Diday (2000) presented an approach to fitting a linear regression model on interval-valued data based on the information given by the midpoints of the intervals. Lima Neto and De Carvalho (2008) improved the former approach presenting a method based on the information given by the midpoints and ranges of the intervals. However, none of these regression models are able to guarantee that the predicted values of the lower bounds will be lower or equal than the predicted values of the upper bounds of the independent variable. In order to guarantee this kind of constrains, we considerer here constrained linear regression models to interval-valued data based on the information given by the midpoints and ranges of the intervals.
|
