1. Title, Journal and Authors

Title: Bayesian networks in environmental modelling

Journal: Environmental Modelling & Software

Authors: P.A. Aguilera, A. Fernández, R. Fernández, R. Rumí b, A. Salmerón

 

 

2. Summary

​This paper was written to explore the current status of Bayesian network(BN) in environmental engineering. This study also deals with their pros and con, availability, and analysis method. ISI Web of Knowledge related to Environmental Sciences, from January 1990 to December 2010, used to search.

 

BNs in environmental modeling, there are some features:

1) BNs are able to use expert opinions constructively when modeling a problem or system.

2) Although input data contain missing values, BNs are still operating and perform the proper predictions.

3) BNs use approximate algorithms posterior probabilities for accurate inference.

With these features, even though we do not know the value of every variable, BNs can predict the value of the response variable in Regression models. Thus, BNs return complete probability distribution for the target variable. The database used in BNs is divided into two different sets, the training set for training the model and the test set for test the model. This approach can avoid overfitting. After the test, Cross Validation and Leave-one-out were used for verification of the model.

 

When referring to ISI Web of Knowledge data, BNs are scarcely applied as a technique in current environmental studies. Most papers use BNs in the model for inference in these circumstances. However, it is a benefit that BN models can adapt incomplete data or discretised continuous variables and reflect expert opinions. Based on these advantages, This study present:

1) A development for an algorithm model that processes continuous and hybrid data in the same network environment would be useful.

2) To use BNs frequently in environmental studies when environmental studies are sometimes obliged to work with incomplete data.

3) Choose a validation technique the best fits the objective of the model.

4) Incorporate expert knowledge for properly using in a BN.

5) Standardize the language of BN modeling.

​

​

3. Originality and Creativity

​

​

4. Application to research

​

​

5. Contact

Suki Park / Intern student

Environmental Systems Engineering Lab.

School of Earth Sciences and Environmental EngineeringGwangju Institute of Science and Technology

1 Oryong-dong Buk-gu Gwangju, 500-712, Korea

Phone : +82-10-9082-5579

E-mail : flskdpffps@naver.com​