Committee: Dejing Dou (chair), Stephen Fickas, Daniel Lowd, Allen Malony, Christopher Wilson
Directed Research Project(Jan 2014)
Keywords: Data Mining; Association Mining; Quantitative Data
The traditional association mining focuses on discovering frequent patterns from the categorical data, such as the supermarket transaction data. The quantitative association mining (QAM) is a nature extension of the traditional association mining. It refers to the task of discovering association rules from quantitative data instead of from categorical data. The discrepancies between the two types of data lead to different analytic methods and mining algorithms. Several properties and interestingness measures that play important roles in the traditional association mining do not apply anymore in the quantitative situation. In this paper, we propose two quantitative association mining algorithms from the bottom up and heuristic search perspectives respectively. They take two new interestingness measures, density and correlation, which are better fits for the quantitative situation. The algorithms can find strong correlated intervals in a generally less correlated environment. Experiment results from neuroscience and health social network data validate the feasibility of our algorithms.