A New Method for Rule Finding Via Bootstrapped Confidence Intervals
Norman Matloff · 2008
Association rule discovery in large data sets is vulnerable to producing excessive false positives, due to the multiple inference effect. Analytical results presented here indicate that Bonferonni-based solutions to this problem may have inherent limitations. Thus the paper proposes a new approach to this problem, based on simultaneous confidence intervals, computed via a novel use of the statistical bootstrap tool. The proposal here differs markedly from previous bootstrap/resampling approaches, not only in function but also in basic goal, which is to enable much more active participation by domain experts. The new method is computationally intensive, but another analytical result presented here has implications for reducing the amount of computation.