As the technologies to collect and store data advances, enumeration of statistically significant patterns is becoming a major challenge, but rigorous statistical evaluation of such patterns has recently been initiated. A limitation of the currently available methods is that they only deal with binary response labels, and thresholding is necessary when dealing with continuous response labels. We present an algorithm that can identify statistically significant itemsets in terms of Wald test based

more »

... on continuous response labels. In computational experiments using synthetic data, we demonstrate that the proposed algorithm is equipped with superior statistically power compared to Bonferroni's method, while correctly controlling Family-Wise Error Rate.