Data-Driven Adaptive Selection of Rules Quality Measures for Improving the Rules Induction Algorithm

Rule Quality Measures Settings in a Sequential Covering Rule Induction Algorithm - an Empirical Approach

“…The quality measure applied in decision rule induction is significant for the quality of an output set of rules. This is confirmed by empirical research [32,20,21,23,25,26].…”

“…In addition, they demonstrate that, as the parameters change, certain measures begin to behave similarly to others (such an analysis was conducted for m-estimates and Klösgen measures). In our articles [26,33], in turn, we are analysing the measures which do not contain parameters and we demonstrate their efficiency on a data set similar to the one quoted in Fürnkranz's work. However, we use an algorithm which generates the rule set instead of a rule list.…”

“…Our range of interest includes objective rule quality measures, also known as evaluation or search heuristics [22,28,29]. At present, in the literature over 50 objective measures can be found [20,22,15,23,28,24,26,29]. A lot of measures used for evaluation of decision (classification) rules have wider significance -they function also as rule interestingness (attractiveness) measures [15,30,16,17,31].…”

“…The selection is a result of a previous experiments -in already quoted works [26,33], about 30 measures were analysed with respect to their efficiency. To check the efficiency, the measures were used in the rule induction algorithm -the same measure was used at each induction stage and during the resolution of classification conflicts.…”

“…If the analysis objective is to create a classification system that uses an interpretable data model, application of sequential covering rule induction algorithms is the most sensible solution. The quality of the rule set obtained by the covering algorithm depends on the quality measure [20,1,21,22,23,24,25,26] used in the growing and pruning phases. The used quality measure is one of the factors affecting the classification accuracy, the number of rules induced and their other characteristics (e.g.…”