FLASH FLOOD PREDICTION IN SELANGOR USING DATA MINING TECHNIQUES

Abstract

Flash floods are one of the most severe natural disasters, which pose a serious threat to infrastructure and human life, especially those in urban areas. As Selangor is one of Malaysia’s most developed and progressive states, the occurrence of flash floods would have a serious impact on the economy and the survival of the people. Hence, the aim of this study was to predict the possibility of flash floods in Selangor by using data mining techniques, specifically logistic regression and artificial neural network. This study proposed six factors, namely, water level, rainfall, durations, weather, minimum temperature, and maximum temperature to be utilised in the flash flood prediction model. The prediction model was constructed based on data gathered from 32 locations in the state of Selangor from June 2020 till March 2021. The performance of the model was compared using the area under the receiver operating characteristic. The findings of this study could be considered as an alternative scientific tool for flash flood prediction and may help to effectively monitor flash flood occurrences in urban areas.