A STUDY ON THE METHOD OF MAKING OCCURRENCE AND NONOCCURRENCE RULES OF DEBRIS FLOW THAT USES SVM AND ROUGH SET

Abstract

  In this study, we construct a rule which can explain many debris flow disasters from gathered data about the debris flow disaster in Hiroshima on June 29, 1999. The analytical technique is to combine Support Vector Machine, excellent in evaluation of the danger, with rough set that can clarify disaster factors for rules. First, we perform evaluation of the danger using Support Vector Machine. From the result, especially safe data and dangerous data are extracted from the database. In the next process, we make typical rules by using rough set with area for expanded rules. By using this method, we could explain debris flow torrents with a few rules from which expanded rules could be extracted.