Quaternary deposit landslides are a representative type of geological disasters in the Three Gorges area. Landslide medium–long-term forecast results offer an important and effective measure to prevent landslide disasters and help guarantee operation security of the reservoir. We applied data mining methods into landslide spatio-temporal data analysis to discover the cause mechanisms, triggered factors and deformation laws, as well as provided medium–long-term forecast criteria for Quaternary deposit landslides in Zigui County in the Three Gorges. We showed that data mining methods used in this study offer an interesting and useful vehicle supporting analysis of a huge amount of landslide spatio-temporal data. The established forecast model exhibits high accuracy, while the discovered forecast criteria come with high confidence and representation capabilities.