Adaptive hybrid ANFIS-PSO and ANFIS-GA approach for occupational risk prediction

Abstract

This study attempted to optimize the adaptive neuro-fuzzy inference system (ANFIS) using particle swarm optimization (PSO) and a genetic algorithm (GA) for calculating occupational risk. Numerous studies have shown that the ANFIS is a good approach for predicting engineering problems. However, it is not well investigated in the area of risk assessment. The proposed techniques were evaluated using various statistical indices, i.e., mean absolute error (MAE) and root mean square error (rmse), to characterize their performance. To test the prediction performance of the proposed technique, a comparison with three well-known machine learning approaches, i.e., artificial neural network (ANN), logistic regression (LR) and support vector machine (SVM), was conducted. The obtained results indicate that ANFIS-PSO achieved better prediction performance for both the training and testing phases. Furthermore, the comparative analysis showed that the proposed model is competitive and suitable for occupational risk prediction.