Integrating WLI fuzzy clustering with grey neural network for missing data imputation

摘要

This paper proposes a novel approach, grey neural network (GNN) that is composed of Levenberg-Marquardt neural network and grey wolf optimiser. The WLI fuzzy clustering mechanism predicts the data by clustering the data into groups, and the neural network trains the missing attribute in the dataset. The Levenberg-Marquardt neural network is trained based on the grey wolf optimiser that determines the optimal weight. Finally, the two imputed values are combined significantly to impute the data where the missing data occurs. Experimentation using the medical dataset proves the accuracy of the proposed hybrid model and the results of the proposed GNN are compared with the existing methods like KNN, WLI and GWLMN. The proposed method exhibits a good efficiency with minimum values of MSE and RMSE compared to the existing methods. This method also attains a minimum RMSE of 0.11 which ensures the efficient data imputation.