A novel hybrid supervised machine learning model for real-time risk assessment of floods using concepts of big data

Risk assessment (RA) modelling refers to combinatorial development of identification and assessment of the potential for the occurrence of an event that causes a negative impact on an entity of interest. With recent advances in data acquisition and archival methods, concepts of big data have been a great boon to RA development. It is primarily due to the fact that the accuracy of RA relies on the volume of historical data analysed. Based on this, a RA model is designed as a hybrid model using differential evolution and an adaptive neuro-fuzzy inference system to assess risk in real-time. The performance ability of the proposed hybrid model is compared with conventional ANFIS and neural network models by analysing the rainfall status in India. Data from the expert systems are collected by analysing various case study areas from India to validate the performance of the proposed hybrid system. The proposed model performance is validated through parameters like precision, recall, f1-score and accuracy. With maximum accuracy of 94.65% proposed model attains better performance than conventional approaches.