Flood Prediction in Nigeria Using Ensemble Machine Learning Techniques


  • K. R. Oloruntoba Department of Computer Science, Federal University Lokoja
  • K. Taiwo
  • J. B. Agbogun


Flood prediction, Machine Learning, Ensemble techniques, Bagging, Boosting


Flooding is the most frequent and destructive natural catastrophe that may happen anywhere in the globe. The frequency and severity of flooding events have increased worldwide in recent years due to climate change and human activity. Flooding has caused widespread death and devastation of property, farms, and vegetation in several emerging African nations, including Nigeria, and has forced the relocation of many more. Flooding has been Nigeria's most common natural disaster during the last decade. Modern machine learning methods have shown great promise for improving flood prediction. The optimum machine learning algorithm for flood prediction is a matter of debate. To reduce the harm caused by floods, finding better ways to anticipate their occurrence is crucial. In this paper, 7 machine learning algorithms (SVM, CART, KNN, GLMNET, LG, LDA and NB) were initially applied on the default dataset. The results reveal fair accuracy (over 60%) and kappa values (< 0.4). The same set of ML algorithms were again applied on the transformed dataset using boxcox transformation technique; the accuracy and kappa values improved but not significantly. Finally, Models for predicting floods were implemented using five different ensemble algorithms: Bagged CART (BAG), Random Forest (RF), Stochastic Gradient Boosting (GBM), Extreme Gradient Boost (XG Boost), and C5.0 (C50). Compared to the other three models, the performance of RF (AUC = 0.93) and BAG (AUC = 0.92) indicated superior accuracy and Kappa.