Abstract: Understanding the causes of landslides from real-time monitoring data is important for identifying triggering mechanisms and potential hazard areas and formulating mitigation measures in a timely manner. However, due to the large amount and diverse sources of monitoring data, the conventional data processing methods can hardly extract useful information from huge monitoring data so as to make a correct evaluation of landslide deformation behaviors and evolution trends. This paper introduces the two-step clustering and association rule analysis methods in the classical data mining methods, proposes the data mining process of association analysis of slope deformation behavior, takes the Xinpu landslide in the Three Gorges reservoir region of Yangtze River as an example, and carries out the association analysis of slope displacement velocity under the influence of reservoir water level and rainfall. The results show that the landslide deformation in the reservoir area is influenced by multiple factors such as the elevation of reservoir water level, reservoir water level fluctuation velocity and rainfall intensity. Water level decline and strong rainfall are closely related to landslide deformation. There are differences in deformation influencing factors at different spatial locations of the landslide. The influence level of reservoir water level fluctuation decreases and the influence level of rainfall intensity increases from front part to rear part. The data mining method can be used to analyze the influencing factors of landslide deformation. It can be used to analyze the influencing factors of landslide deformation, and the comparison with the measured data verifies the reliability of the rules. These results are important for the analysis of the causes of landslide disasters under the massive monitoring data.