Associative classification is becoming a promising alternative to classical machine learning algorithms. It is a hybrid technique that combines supervised and unsupervised data mining algorithms and builds classifiers from association rules' models. The aim of this work is to apply these associative classifiers to improve estimation precision in the project management area where data sparsity involves a major drawback. Moreover, in this application domain, most of the attributes are continuous; therefore, they must be discretized before generating the rules. The discretization procedure has a significant effect on the quality of the induced rules as well as on the precision of the classifiers built from them. In this paper, a multivariate supervised discretization method is proposed, which takes into account the predictive purpose of the association rules.
展开▼
