Development of a Human-Elephant Conflict (HEC) Vulnerability Model in Pidie District, Aceh Province.

Abstract

This paper examines and develops a model of elephant conflict vulnerability using both parametric and non-parametric approaches, integrating several anthropogenic and geo-biophysical spatial variables in Pidie Regency, Aceh Province. The parametric approach involves building a logistic regression model, while the non-parametric approach uses a decision tree machine learning. The purpose of this study is to develop a spatial model HEC vulnerability using spatial data of anthropogenic and biophysical factors. This vulnerability model is built using variables related to habitat, elephant feed sources, and disturbances, including elevation, slope, proximity to water bodies, proximity to plantations, proximity to forest boundaries, proximity to roads, and proximity to settlements. Data on the location of HEC used as response variables are obtained from previous conflict incidents, official reports from government agencies, and community reports that are reliable. This study reveals that road proximity, slope, and plantation proximity are three critical predictor variables for assessing the level of HEC vulnerability. Model evaluation using the ROC- AUC approach demonstrated good performance, with an AUC value of 0.878 and an overall accuracy of 82.54%. This study concluded that conflict vulnerability can be developed to accurately mitigate HEC, which may be used to inform the development of mitigation strategies.