Research on the Evaluation of Association Rules Based on"Integrity-Reliability"
In mining association rules,less attention was paid to the relationship between rules by the evaluation indicators.In view of this,two new evaluation indexes were proposed.The definitions and calculation methods for integrity and reliability were introduced.A"com-pleteness-reliability"evaluation system was constructed to determine the level of completeness of the association relationship reflected by the association rule within the data set,and to quantitatively evaluate the reliability of rule results.By validating the evaluation effect in the associ-ation rule mining results of lightning data,the results indicated that the"completeness-reliability"evaluation system assessed association rules at multiple levels,effectively achieved loophole detection and improvement in association rule mining results,enhanced the accuracy and usability of rules,and provided strong decision support for lightning disaster prediction and prevention work.
association rulesthe"integrity-reliability"evaluation systemrule result evaluationlightning activity prediction
万方数据
