Knowledge discovery by inspection

Abstract Given the enormous size of many business databases, algorithms for knowledge discovery can often be applied to only a sample of the original data. Other methods used to improve efficiency include focusing on a restricted class of rules such as exact rules, or limiting the number of conditions in the discovered rules. It is shown that simple exact rules can often be discovered by visual inspection of frequency tables. An efficient algorithm for rule discovery by inspection is presented. The discovered rules include all exact rules with one or two conditions.