Publication

Abstract : The changes in climatic conditions begun to impact various regions around the world. Flood forecasting is a fundamentally challenging task because of its uncertain nature. Therefore, flood forecasting has been a key research area in the field of hydrology. This study introduces a machine learning approach that make use of meta classifiers—RandomSubSpace, LogitBoost and tree classifiers—RandomForest and REPTree for prediction. A combined approach of classifiers is selected for the forecasting of flood. The use of Tree classifiers along with Meta classifiers produce a better result to analyse the weather conditions that trigger the flood. The experimental results show that the Error rate can be effectively reduced by formulating the Tree classifiers with Meta classifiers and thereby the forecast accuracy can be improved.