Support Vector Regression on GIS and Map Data for Landslide Susceptibility of Laguna, Philippines
Abstract— Given the province of Laguna’s geographical situation, it is prone to different kinds of natural hazards. Landslide controlling factors are present in the area. This study attempted to create a landslide susceptibility mapping using an integrated weight index model by applying analytical hierarchy process (AHP) and frequency ratio (FR) on GIS and map based environmental data acquired from different government and nongovernment organizations. The features are extracted and merged using QGIS and Python scripts. The susceptibility model is then fitted into a Support Vector Regression model to find the correlation of the extracted factors to the respective landslide susceptibility ratings. A web-application was also develop to enable interaction with the resulting models.
Objectives
The study, in general, aims to apply Support Vector Regression on a workflow developed for creating landslide susceptibility maps.
1) To develop a computational workflow for creating landslide susceptibility maps,
2) To create the same susceptibility maps using less factors using Support Vector Regression(SVR),
3) To assess the effectiveness of SVR on modeling landslide susceptibility using less factors, and
4) To develop a web application that will visualize the resulting susceptibility maps and results.