ISSUES ON PRACTICAL APPLICATIONS OF SLOPE STABILITY EVALUATION MODELS AND A PROPOSAL OF AN EVALUATION INDEX

Abstract

This paper illustrates issues on practical applications of slope stability evaluation models, and has proposed an ambiguity index for evaluating results derived from these models. Quantitative models generally construct the relationship between past slope failures and causal factors, and “hazardous or non-hazardous” areas are delineated on the slope stability evaluation map, respectively. Based on those classified results, Quantity of Acquired information (QA) has been formulated based on the concept of “conditional entropy.” Through the statistical measure of correlation ratio, we conclude that the proposed index of QA is superior, for selecting the models, to the general index of “success rate” with respect to training data sets (i.e. outside variables).