TY - GEN
T1 - Water Quality Index Analysis and Prediction
T2 - 29th International Conference on Information Modeling and Knowledge Bases, EJC 2019
AU - Chawakitchareon, Petchporn
AU - Thalheim, Bernhard
AU - Kiyoki, Yasushi
N1 - Funding Information:
This research was supported in part by Faculty of Engineering Chulalongkorn University under the enhancement of engineering academic collaboration with the federal Republic of Germany Project. This research is also in part support by GESL program, Keio University, Japan. The authors would like to thank the Bangkok Metropolitan Office, Thailand for the data provided.
Publisher Copyright:
© 2020 The authors and IOS Press. All rights reserved.
PY - 2019/12/13
Y1 - 2019/12/13
N2 - This paper presents a comparison of prediction methods for a water quality index (WQI) that is used for classification of water quality in rivers or canals. In this work, we consider the water quality index of two canals namely Phadung Krung Kasem Canal and Saen Saep Canal, Bangkok, Thailand as a case study. We compare results from M5P, M5Rules, REPTree with results from multilayer perceptron. The models employ five input variables including dissolved oxygen (DO), biological oxygen demand (BOD), ammonia nitrogen (NH3-N), Fecal Coliform bacteria (FCB) and Total Coliform bacteria (TCB) which were measured in the canals. The data in this research had been collected from Bangkok Metropolitan Authority, Thailand from 1 January 2007 to 31 November 2017. The total number of data is 2,000 records. The 10-fold cross validation method is used for evaluation of prediction models. It allows to determine the most effective method. Our experimental results show that the REPTree method yielded the highest accuracy to predict water quality index compared to other methods proposed in this paper.
AB - This paper presents a comparison of prediction methods for a water quality index (WQI) that is used for classification of water quality in rivers or canals. In this work, we consider the water quality index of two canals namely Phadung Krung Kasem Canal and Saen Saep Canal, Bangkok, Thailand as a case study. We compare results from M5P, M5Rules, REPTree with results from multilayer perceptron. The models employ five input variables including dissolved oxygen (DO), biological oxygen demand (BOD), ammonia nitrogen (NH3-N), Fecal Coliform bacteria (FCB) and Total Coliform bacteria (TCB) which were measured in the canals. The data in this research had been collected from Bangkok Metropolitan Authority, Thailand from 1 January 2007 to 31 November 2017. The total number of data is 2,000 records. The 10-fold cross validation method is used for evaluation of prediction models. It allows to determine the most effective method. Our experimental results show that the REPTree method yielded the highest accuracy to predict water quality index compared to other methods proposed in this paper.
KW - Artificial neural network
KW - Data mining
KW - M5p
KW - M5rules
KW - Machine learning
KW - Mlp
KW - Multilayer perceptron
KW - Reptree
KW - Water quality index
KW - Weka
KW - Wqi
UR - http://www.scopus.com/inward/record.url?scp=85082518972&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85082518972&partnerID=8YFLogxK
U2 - 10.3233/FAIA200028
DO - 10.3233/FAIA200028
M3 - Conference contribution
AN - SCOPUS:85082518972
T3 - Frontiers in Artificial Intelligence and Applications
SP - 419
EP - 429
BT - Information Modelling and Knowledge Bases XXXI
A2 - Dahanayake, Ajantha
A2 - Huiskonen, Janne
A2 - Kiyoki, Yasushi
A2 - Thalheim, Bernhard
A2 - Jaakkola, Hannu
A2 - Yoshida, Naofumi
PB - IOS Press
Y2 - 3 June 2019 through 7 June 2019
ER -