Page 235 - Contributed Paper Session (CPS) - Volume 2

CPS1845 Devni P.S. et al.

Bayesian network for risk management of
earthquake damage using R and Genie
1
1
Devni Prima Sari 1,2,* ; Dedi Rosadi , Adhitya Ronnie Effendie , Danardono 1
1 Department of Mathematics, Universitas Gadjah Mada, Yogyakarta, Indonesia
2 Department of Mathematics, Universitas Negeri Padang, Padang, Indonesia

Abstract
The occurrence of an earthquake in an area is often unexpected. Disaster risk
management is needed to minimize disaster risk. One way to reduce the
impact of disasters is to know the level of risk of damage to buildings in an
area. The level of risk of damage to buildings can we value the variables that
participate in determining the size or smallness of a risk. The tool we will use
to deduce a causal relationship between variables is called the Bayesian
Network. The formation of the Bayesian Network model is a complex and
time-consuming task. It is difficult to get a complete and consistent model. To
simplify work while struggling with BN, we use the famous R package with BN
called "bnlearn" and GeNIe Software. The process of forming BN is faster and
more efficient with the role of bnlearn and GeNIe.

Keywords
Earthquake; Risk; Bayesian Network; bnlearn; GeNIe

1. Introduction
Earthquake disasters that hit a country often occur unexpectedly, so that
people who are in the location of the disaster, do not have time to anticipate
the prevention of the disaster. Earthquakes certainly cause destruction, also
cause suffering and loss both for society and the country. Therefore, disaster
risk management is needed to minimize the risk of disaster. The problem that
arises is that there is still a lack of knowledge of citizens about what a disaster
is, how to anticipate and reduce the impact of disasters. One way to reduce
the impact of disasters is to know the level of risk of damage to buildings in
an area. The level of risk of damage to buildings can we value from the
variables that influence it.
The tool we will use to infer the relationship of the causal variable from
correlation is called a Probabilistic Graphic Model (PGM). PGM is a graph that
encodes a causal relationship between events. Each node represents a random
variable, and arrows represent a dependency relationship between variables.
PGM with directional and non-cycle edges is specifically called a Bayesian
Network (BN), and that is the type of PGM that I will focus on. BN has been
developed in various fields, including in medical (Flores, Nicholson, Brunskill,
Korb, & Mascaro, 2011), financial (Neil, Marquez, & Fenton, 2008). As well as

224 | I S I W S C 2 0 1 9