Flood Endangered Area Classification Using the K-Nearest Neighbour Algorithm

Abstract

Preparing for the uncertainty of life is one aspect of the human existence that cannot be over emphasized. With the growth of technology especially the sophisticated nature of data mining and machine learning algorithms, these uncertainties can be predicted, planned and prepared for using existing variables and computer methodologies. The achievements and accomplishments of big data analytics over the past decade in diverse areas called for its implementation in meteorological and space data. Notably, enhancement of the proper management of life’s uncertainties when they eventually occur. This research work focuses on the classification of areas within the Nigerian Geographical territory that are prone to flood using the K-nearest neighbour Algorithm as a classifier. Data from Nigeria Meteorological Agency (NiMET) on seasonal rainfall prediction and temperature of different stations and cities for over three (3) years (2014-2017) was used as a dataset which was trained and classified with the k-Nearest Neighbour algorithm of machine learning. Results showed that some areas are prone to flood considering the historic data of both rainfall and temperature.