Disaster mitigation solutions with Bayesian network

Abstract

Natural disasters are events that threaten and disrupt the lives and livelihoods of people caused by natural factors. Indonesia is a country prone to natural disasters. This is triggered by its geographical position flanked by two large oceans and geologically at the confluence of the three main plates. One way to reduce the impact of natural disasters is to mitigate hazards. Mitigation will reduce the negative impact caused by disasters, especially for residents. It can also be a guideline for development planning. The way of mitigation efforts can be done by introducing and monitoring disaster risk. For example, to observe what factors affect the level of building damage caused by a disaster. We can do this with the Bayesian Network approach because this approach provides flexibility in seeing relationships between variables and adding new variables based on expert analysis. These advantages are very supportive related to cases of natural disasters; sometimes, there are often developments in variables that affect the level of damage in the field. The first step in the approach is to form a structure. In this study, we conducted two types of structure formation, namely using the Naive Bayes algorithm and expert opinion. From these two methods, the creation of a structure based on expert opinion is more accurate.