Abstract: In this paper is proposed a new method for mining quantitative association rules by means of principal component analysis. In contrast to traditional Boolean data matrix, our algorithm is based on quantitative datab ase, which contains some value knowledge for us. The experiment illustrated that the algorithm also works well on canal binary data matrices. Based on the idea of principal component analysis (PCA) discovering the principal component as quantitative association, the ratio item set is definited as well. In additional, it is shown that for the treatment of the complexity of time and space the method is more effective than traditional methods.

Key words: association rules, principal component analysis, singul ar value decomposition, ratio item set