Obtención de Reglas Significativas e Interpretables para Descubrimiento de Subgrupos (Obtaining Significant and Interpretable Rules for Subgroup Discovery Tasks)

Juan Luis Domínguez Olmedo (juan.dominguez@dti.uhu.es)1, Jacinto Mata Vázquez (jacinto.mata@dti.uhu.es)1


1Universidad de Huelva

This paper appears in: Revista IEEE América Latina

Publication Date: Oct. 2017
Volume: 15,   Issue: 10 
ISSN: 1548-0992


Abstract:
In this work, a deterministic method to obtain rules for subgroup discovery tasks is described. It does not employ a previous discretization for the numeric attributes, but obtains conditions for the rules dynamically. To select the final rules, the equivalent sum of sensitivity and specificity has been used. An experimental study and appropriate statistical tests were performed, in a comparison with the classic deterministic algorithms APRIORI-SD and CN2-SD. The proposed method behaved very well, obtaining the best results in six of the eight pair-wise comparisons. From the results obtained, the described approach can be seen as a suitable deterministic way of extracting rules for subgroup discovery problems.

Index Terms:
data mining, subgroup discovery, deterministic algorithms   


Documents that cite this document
This function is not implemented yet.


[PDF Full-Text (378)]