Multilayer perceptron with Cuckoo search in water level prediction for flood forecasting

Abstract

The feed forward multilayer perceptron (MLP) with the Cuckoo search (CS) algorithm, called CS-MLP is implemented to predict 7-hours-ahead water level of the Ping river at the downtown area of Chiang Mai, Thailand. The CS-MLP model prediction performance is compared with the regular multilayer perceptron (MLP) and the results from the previous work. The CS-MLP is the best among them with the mean absolute error on the blind test data set of 6.836 cm.