This paper proposes a novel method to improve the accuracy of tropical cyclone (TC) rainfall prediction in the Western North Pacific (WNP) by complementing the Fuzzy C Means (FCM) approach for identifying similar TCs. filters to refine the list of TCs identified by FCM, which are then used as ensemble members for rainfall averaging. Unlike traditional approaches that use a fixed number of members for all TC targets, our method produces a varying number of members depending on the similarity of the TCs. rely on FCM, indicating its potential to enhance the accuracy of TC rainfall prediction in the WNP region.