SHapley Additive exPlanations (SHAP) for Landslide Susceptibility Models: Shedding Light on Explainable AI

The paper attempts to create a new technique for assessing the current state of land degradation. Assessment of land degradation is difficult, because it includes a complex process. This assessment, using RS and GIS seems to be more realistic in finding the degree of degradation, because it is more related to its impact on land productivity. It is hoped that this attempt, which is the first attempt of its kind in the world, will be found applicable for other regions.

The main contributions of this manuscript are • A novel and easy to use ArcMap tool is proposed for bivariate statistical analysis model implementation. • The developed tool facilitated and sped up the BSA modelling process of frequency ratio, weights of evidence, and evidential belief function. • This tool classifies the variables based on the acquired weights automatically. • All three methods were tested on a study area and landslide susceptibility maps were produced.