Landslide Hazard Assessment Based On Uncertain Data Mining Classification Method

China is a country of frequent occurrence of geological hazards, especially the landslide disaster, in case of rain, in some places it is prone to landslides, causing harm to people’s lives,property losses. Therefore, how efficient regional landslide hazard prevention is a very important issue.Data mining is a new data analysis techniques, which can learn by classification method to extract the rules, and thus to predict the unknown. However landslide hazard prediction is affected by many factors, including rainfall and other uncertainties which are difficult to obtain data and effectively deal with.To improve the landslide hazard prediction accuracy,according to the theory of landslide disasters and decision tree classification principle in data mining, this paper proposes two methods: uncertainty C4.5 decision tree algorithm and uncertainty fuzzy ID3 decision tree algorithm, respectively, to predict landslide hazard.This paper firstly introduces the landslide disasters and data mining theory, which is the theoretical foundation for the later chapters. Then introduces the high precision uncertainty C4.5 decision tree algorithm, and apply it to an instance for testing results. Based on the traditional fuzzy ID3 algorithm, using the integral thinking to deduce fuzzy processing methods for uncertain data and fuzzy ID3 algorithm, we propose a new uncertain fuzzy ID3 algorithm and establish uncertain fuzzy ID3 decision tree model, and predict landslide hazard region, which can ensure the accuracy of the algorithm on the basis of simplifying the complexity of the algorithm.Finally, by comparing and analysing the traditional algorithm and the algorithm proposed in this paper, this paper tries to find out the best fit situation of each algorithm and provide references for algorithm selecting.