DATA MINING

Correlation Computing. A large body of association mining work in data mining was motivated by the difficulty of efficiently identifying the highly correlated objects in huge data volumes. This has led to the use of alternative interest measures, e.g. support and confidence, despite the lack of precise relationship between these new interest measures and statistical correlation measures. However, this approach is not effective for discovering potentially interesting patterns at low levels of support. Also, it tends to generate too many spurious patterns involving objects which are poorly correlated. There is a clear desire for characterizing the relationship between statistical correlation measures and new association measures towards the development of a science of data mining. Along this line, we provided a precise relation between Pearson's correlation coefficient and the support measure. We also proposed an efficient algorithm called TAPER to identify highly-correlated pairs of..