Comparative analysis of machine learning and multi-criteria decision making techniques for landslide susceptibility mapping of Muzaffarabad district

Landslides are natural disasters deliberated as the most destructive among the others considered. Using the Muzaffarabad as a case study, this work compares the performance of three conventional Machine Learning (ML) techniques, namely Logistic Regression (LGR), Linear Regression (LR), Support Vector Machine (SVM), and two Multi-Criteria Decision Making (MCDM) techniques, namely Analytical Hierarchy Process (AHP) and Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) for the susceptibility mapping of landslides. Most of these techniques have been used in the region of Northern Pakistan before for the same purpose. However, this study for landslide susceptibility assessment compares the performance of various techniques and provides additional insights into the factors used by adopting multicollinearity analysis. Landslide-inducing factors considered in this research are lithology, slope, flow direction, fault lines, aspect, elevation, curvature, earthquakes, plan curvature, precipitation, profile curvature, Normalized Difference Water Index (NDWI), Normalized Difference Vegetation Index (NDVI), roads, and waterways. Results show that SVM performs better than LGR and LR among ML models. On the other hand, the performance of AHP was better than TOPSIS. All the models rank slope, precipitation, elevation, lithology, NDWI, and flow direction as the top three most imperative landslide-inducing factors. Results show 80% accuracy in Landslide Susceptibility Maps (LSMs) from ML techniques. The accuracy of the produced map from the AHP model is 80%, but for TOPSIS, it is less (78%). In disaster planning, the produced LSMs can significantly help the decision-makers, town planners, and local management take necessary measures to decrease the loss of life and assets.

Abstract licence: CC BY

94 more references, whose wording the paper’s licence keeps private.

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.