Abstract:

By using 10-year monitoring data of geological disaster in Jilin Province, we focused on risk assessment and partition of susceptible area of geological disaster with statistics methods. Firstly, hierachical cluster analysis was improved by introducing into weight coefficient based on the importance of different factors, and the accuracy of partition was tested to achieve 94.87% by discriminant analysis. Secondly, a threshold regressive model whose accuracy got to 99.996 2% was used to learn the forms of disaster caused by all kinds of factors and master their features and differences of disaster formation under numerical values of different sizes of different factors. The high level of accuracy of the combination application of weighted hierachical cluster analysis with threshold regressive model showed it was proper methods to resolve the issue of risk assessment and our partition of susceptible area of geological disaster was precise.

Key words: risk assessment, geological disaster, hierachical cluster, threshold regressive, weight