Predicting Flood in Perlis Using Ant Colony Optimization
Authors:
Open Access
- 7 June 2017
- conference paper
- Published by IOP Publishing in Journal of Physics: Conference Series
- Vol. 855 (1) , 012040
Abstract
Flood forecasting is widely being studied in order to reduce the effect of flood such as loss of property, loss of life and contamination of water supply. Usually flood occurs due to continuous heavy rainfall. This study used a variant of Ant Colony Optimization (ACO) algorithm named the Ant-Miner to develop the classification prediction model to predict flood. However, since Ant-Miner only accept discrete data, while rainfall data is a time series data, a pre-processing steps is needed to discretize the rainfall data initially. This study used a technique called the Symbolic Aggregate Approximation (SAX) to convert the rainfall time series data into discrete data. As an addition, Simple K-Means algorithm was used to cluster the data produced by SAX. The findings show that the predictive accuracy of the classification prediction model is more than 80%.Keywords
This publication has 10 references indexed in Scilit:
- Fuzzy Spatial Forecasting Model of Rainfall Distribution for Flood Early Warning2016
- Modelling of Reservoir Water Release Decision Using Neural Network and Temporal Pattern of Reservoir Water Level2014
- Data Mining: Practical Machine Learning Tools and Techniques2011
- The WEKA data mining softwareACM SIGKDD Explorations Newsletter, 2009
- Ant Colony Optimization2004
- A symbolic representation of time series, with implications for streaming algorithms2003
- Data mining with an ant colony optimization algorithmIEEE Transactions on Evolutionary Computation, 2002
- An Ant Colony Algorithm for Classification Rule Discovery2002
- Locally adaptive dimensionality reduction for indexing large time series databasesACM SIGMOD Record, 2001
- Ant system: optimization by a colony of cooperating agentsIEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), 1996