Measuring interestingness of discovered skewed patterns in data cubes

This paper describes a methodology of OLAP cube navigation to identify interesting surprises by using a skewness based approach. Three different measures of interestingness of navigation rules are proposed. The navigation rules are examined for their interestingness in terms of their expectedness of skewness from neighborhood rules. A novel Axis Shift Theory (AST) to determine interesting navigation paths is presented along with an attribute influence approach for generalization of rules, which measures the interestingness of dimensional attributes and their relative influence on navigation paths. Detailed examples and extensive experiments demonstrate the effectiveness of interestingness of navigation rules.