Abstract | ||
---|---|---|
Effective data mining techniques are crucial for analysing and extracting useful information from large datasets. The aim of this research is to investigate different techniques for detecting the variation in association rules over time. Analysing the rules instead of the large datasets allows one to have a better understanding of the overall trend of the entire database. In this paper, we propose a methodology to extract and categorise rules from a time driven database. We develop metrics to identify the level of variation for each rule and the entire rule set over a certain time period. The metrics exploit multi-resolution techniques to improve computational performance. In our experiment, we apply the proposed methods to different sets of simulated data. We demonstrate how simple patterns can be detected in more complex datasets. Additionally, we illustrate how these patterns can be rapidly identified using a graphical representation. |
Year | DOI | Venue |
---|---|---|
2012 | 10.1145/2381716.2381810 | CUBE |
Keywords | Field | DocType |
association rule variation,entire database,different technique,certain time period,entire rule,effective data mining technique,complex datasets,different set,categorise rule,large datasets,association rule,sequence mining,association rules | Data mining,Computer science,Exploit,Association rule learning,Sequential Pattern Mining | Conference |
Citations | PageRank | References |
0 | 0.34 | 11 |
Authors | ||
2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Monica H. Ou | 1 | 15 | 3.65 |
Jérôme Maillot | 2 | 224 | 35.62 |