关键词: 不确定贝叶斯模型, 滑坡, 危险性评价, 数据预处理

Abstract: The traditional Bayesian classification technology can’t obtain the effective parameters which the prediction model needs during constructing the classification model. And the method to describe the inducing factors generating a landslide quantificationally is difficult. Therefore, the uncertain Bayes algorithm is introduced. The uncertain Bayes classification model is constructed by combining the uncertain world model with the naive Bayes classification model. Finally, the goal to effectively describe the character of the rainfall which is an uncertain attribute and improve the accuracy of the landslide hazard prediction is achieved. Experiment over a realistic dataset reveals that the approaches improve the landslide risk evaluation significantly.

Key words: uncertain Bayes model, landslide, hazard assessment, data preprocessing