Abstract
Landslides are one of the most naturally occurring phenomena with the highest human and economic losses around the world, reason for the susceptibility and hazard assessment is a fundamental tool for land use planning. There is a wide range of Artificial Intelligence algorithms in the recent literature with completely different approaches to establish the relationship between the independent variable (predictors) and the dependent variable (landslide inventory). In the present study, a wide range of algorithms were used for the La Miel creek basin, in the Colombian Andes, and the methodology implemented for this type of data-based modeling is presented in detail and step by step. The results obtained show that the assembled boosting models present the best values in terms of performance and predictability. Contrasting with the linear parametric models, pointing out their limitations in modeling complex problems such as landslides.

This work is licensed under a Creative Commons Attribution 4.0 International License.
Copyright (c) 2021 Juan Pablo Ospina-Gutiérrez, Edier Aristizábal