Logo image
搜尋K個最頻繁閉項目集的複合方法
Thesis

搜尋K個最頻繁閉項目集的複合方法

黃偉哲
Masters, 國立清華大學, 資訊工程學系
2004

Abstract

資料探勘 最小支持 頻繁 最低長度 Data Mining Minimum support Frequent Closed Minimal Length
More and more data cause the size of database very large. How to find the data that user interested in is a important task like Data Mining. This paper is about Data Mining. In the paper “A Hybrid Method for Mining Frequent Closed Patterns”, we combine the method about horizontal format and vertical format to improve performance of mining. During 2002, a new task has be presented. Finding the K most frequent closed patterns that its length no smaller than Minimal Length(TFP). The K and Minimal Length are defined by users. The difference with traditional method is users do not need to defined minimum support. Minimum support is rising dynamically during mining. Our method is like “A Hybrid Method for Mining Frequent Closed Patterns”, mining based on intersection of patterns. Because of the constraint “Minimal Length”, we can prove a lot of candidates which its length small than Minimal Length during mining. In general datasets, with increasing Minimal length, our mining is more efficient. In the special datasets, like many columns but less rows, our method outperforms TFP.

Metrics

1 Record Views

Details

Logo image