In this paper we present a new score to determine when two categorical variables are independent. It represents a measure that can be used in classification. It is an interval-valued score that is based on the Heckerman, Geiger, and Chickering's score. We also carry out an empirical comparison with different scores to determine when two binary variables are independent. The others measures that have been considered are: the Bayesian score metric, the Bayesian information criterion (BIC), the p-value of the Chi-square test for independence and the upper entropy score based on imprecise probabilities. For the new score, we find a behaviour that it is more similar to statistical tests from small samples and to Bayesian procedures for large samples. This makes it very appropriate for some concrete types of problems.
Keywords. Independence, statistical tests, Bayesian score, Chi-square test, imprecise Dirichlet model
Paper Download
The paper is availabe in the following formats:
Authors addresses:
Joaquín Abellán
Dpto. Ciencias de la Computaci�n
ETSI Inform�tica
18071 Granada
SPAIN
Serafín Moral
Dpto. Ciencias de la Computación e IA
ETSI Informática
Universidad de Granada
18071 Granada - Spain
E-mail addresses:
Joaquín Abellán | jabellan@decsai.ugr.es |
Serafín Moral | smc@decsai.ugr.es |