Data acquisition and discretization for flood correlation model

• Main Authors: Ahmad Azami, Nor Idayu, Yusoff, Nooraini, Ku-Mahamud, Ku Ruhana
• Format: Article
• Language: English
• Published: 2017
• Subjects: QA76 Computer software
• Online Access: http://repo.uum.edu.my/21721/1/JTAIT%2095%204%202017%20879-889.pdf; http://repo.uum.edu.my/21721/; http://www.jatit.org/volumes/Vol95No4/15Vol95No4.pdf

Flood is among the natural disasters caused by complex factors such as natural, breeding, and environmental. Moreover, the variability of such factors on multiple heterogeneous spatial scales may cause difficulties in finding correlation or association between regions under study.The interaction between these factors may result in the provision of either diverse or repeated information, which can be detrimental to prediction accuracy. Therefore, in this study, a model has been developed to find the association between the factors that cause flood.In particular, a Bayesian Network-based method is proposed to quantify the dependency patterns in spatial data. It has been shown that although many factors may be important with respect to the flood for a particular region, the same factors may not be important for other regions.The probabilistic model has been successfully used in problems in which the dependency between the factors is of interest.Furthermore, the effect of the proposed fuzzy discretization on the association performance has also been investigated. The comparison between different data discretization techniques proved that the proposed method gives a better result with the precision of 0.992, F-measure of 0.980, and receiver operating characteristic of 0.984 for three correlation models, respectively.