A hybrid genetic algorithm for climate input features and neural network parameters selection

The climate input features and neural network parameters highly affect the overall performance of the rainfall prediction models. In this paper, a novel approach is proposed to select the input features and neural network parameters. A new hybrid genetic algorithm that combines natural reproduction and particle swarm optimization characteristics was developed to select the best climate features and network parameters. The developed model was compared against alternative models including climatology and showed a better accuracy. The aggregated time series of the proposed model showed a Root Mean Square Error (RMSE) of 141.67 mm for a location with 3553.00 mm annual average.