Universidad Complutense de Madrid
E-Prints Complutense

Criterion for the selection of an optimum tree with respect to association coefficients obtained from the x2.(Spanish:Criterio de selección de un árbol óptimo según coeficientes de asociación derivados de x2)

Impacto

Downloads

Downloads per month over past year



Cano Sevilla, Francisco Jose and Muntuate del Rio, A. and Pérez Prados, A. (1996) Criterion for the selection of an optimum tree with respect to association coefficients obtained from the x2.(Spanish:Criterio de selección de un árbol óptimo según coeficientes de asociación derivados de x2). Qüestiió, 20 (2). pp. 167-186. ISSN 0210-8054

[img] PDF
Restringido a Repository staff only

1MB

Official URL: http://www.idescat.cat/sort/questiio/questiiopdf/20.2.2.cano.pdf



Abstract

“First we analyze the variation produced in the value of
the contingency coefficient by a pruning process in a decision tree T. Then we define a criterion that linearly combines the contingency coefficient and the simplicity index. Using this criterion, we propose a method for obtaining an optimum tree for each of the distinct values of the parameter � of the linear combination. To select the optimum tree from among those trees, we use the Chuprov
coefficient, depending on the two measures considered for the quality of the tree.”


Item Type:Article
Uncontrolled Keywords:Árboles de decisión; Proceso poda; Coeficiente de contingencia; Coeficiente de Tschuprow; Simplicidad; Árbol óptimo
Subjects:Sciences > Mathematics > Bayesian statistical decision theory
ID Code:20569
Deposited On:21 Mar 2013 14:46
Last Modified:08 Aug 2018 07:42

Origin of downloads

Repository Staff Only: item control page