The hypothesis
space, \(H,\) may contain a huge (possibly infinite) number of functions
. Randomly producing functions
, in the hope of finding one that approximates well the target function \({P}(y|x),\) is not likely to be an efficient strategy. Fortunately, there are countless more efficient strategies for producing approximations
of \({P}(y|x).\) One of these strategies is to directly exploit
... [Show full abstract] relationships, dependencies and associations
between inputs
and outputs
(i.e., classes) (Liu et al. Integrating classification and association rule mining. In: Proceedings of the Conference on Data Mining and Knowledge Discovery (KDD), 1998). Such associations
are usually hidden in the examples in \(S,\) and, when uncovered, they may reveal important aspects concerning the underlying phenomenon that generated these examples (i.e., \({P}(y|x)\)). These aspects can be exploited for the sake of producing only functions
that provide potentially good approximations
of \({P}(y|x)\) (Cheng et al. Discriminative frequent pattern analysis for effective classification. In: Proceedings of the International Conference on Data Engineering (ICDE), 2007; Cheng et al. Direct discriminative pattern mining for effective classification. In: Proceedings of the International Conference on Data Engineering (ICDE), 2008; Fan et al. Direct mining of discriminative and essential frequent patterns via model-based search tree. In: Proceedings of the Conference on Data Mining and Knowledge Discovery (KDD), 2008; Li et al. Efficient classification based on multiple class-association rules. In: Proceedings of the International Conference on Data Mining (ICDM), 2001). This strategy has led to a new family of classification
algorithms
which are often referred to as associative classification algorithms
. The mapping functions
produced by these algorithms
are composed of rules
\(X\rightarrow c_j,\) indicating an association
between \(X,\) which is a set of attribute-values, and a class \(c_j\in y.\) In this chapter, we present and evaluate novel associative classification algorithms
.