The invention discloses a
landslide susceptibility assessment method and
system fusing feature interaction and dynamic weighting, and belongs to the field of
geological disaster assessment, and the method comprises the steps: building an evaluation
index system, and carrying out the factor classification; acquiring
landslide and non-
landslide point sample data, evaluating factor importance by adopting a
random forest to obtain a preliminary weight, and preliminarily evaluating the susceptibility based on an I-RF model; detecting double-factor interaction by using a geographic
detector, screening interaction factors for collaborative enhancement, and optimizing an expansion
index system; retraining the
random forest model, obtaining a new weight containing a
single factor and an
interaction factor, and constructing an I-RF optimization model; for
spatial heterogeneity, sub-regions are divided by using
spatial clustering, and the weight of each sub-region is dynamically corrected according to the sensitivity of a main
control factor, so that self-adaptive adjustment is realized; and calculating susceptibility indexes of the sub-regions and the whole region, and grading to form spatialized and multi-level risk partitions. According to the method, the scientificity and accuracy of evaluation are effectively improved through feature interaction optimization and partition dynamic weighting.