LIU Rui1,2, LI Luyao1,3, YANG Xin1, YANG Yuantao1, YANG Mei1
（1.Key Laboratory of Earth Exploration and Information Technology of Ministry of Education, Chengdu University of Technology, Chengdu, 610059,China;
2.State Key Laboratory of Geohazard Prevention and Geoenvironment Protection, Chengdu University of Technology, Chengdu, 610059 ,China；
3.Chengdu University of Technology, Chengdu, 610059,China）
Abstract: In the process of landslide susceptibility mapping, the performance of the landslide prediction model depends largely on the selection and combination of influencing factors. How to select landslide impact factors to optimize the performance of landslide prediction model is the key to the evaluation of regional landslide susceptibility. This paper proposes an Apriori algorithm which based on the strong correlation analysis, and through landslide strong correlation analysis, the factors with the highest probability of inducing disaster were selected from 15 pre-selected landslide factors. Then use random forest building liability landslide prediction model, the results before and after the optimization of the factors are compared to measure the impact of the algorithm on the model prediction performance. The results show that the landslide prone area obtained by the optimized factors group is more consistent with the actual landslide distribution and the prediction model is more accurate. The Apriori algorithm improves the combination method of landslide impact factors，thereby providing a new idea for the landslide sensitivity prediction method.
Key words: Apriori; factor selection；random forest；landslide susceptibility mapping
EARTH AND ENVIRONMENT Vol.49, No.2, Tot No.340, 2021, Page 198-206