The quality of single classification rules for numerical data can be evaluated by different measures. Common measures are the frequency and the confidence of rules beside others. A problem with these measures is that they are valid for a rule only if an uniform distribution of the data, corresponding to the rule, is assumed. Since this is usually not the case, especially when considering high dimensional data, subrules and their properties should be considered additionally. The frequency and the confidence values of the subrules, summarized in a diagram, give more information about the quality of the rules than the properties of the rules solely. © Springer-Verlag Berlin Heidelberg 2007.
CITATION STYLE
Paetz, J. (2007). Subrule analysis and the frequency-confidence diagram. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 4723 LNCS, pp. 219–228). Springer Verlag. https://doi.org/10.1007/978-3-540-74825-0_20
Mendeley helps you to discover research relevant for your work.