A Novel Soft Computing Hybrid for Data Imputation

We propose a novel 2-stage soft computing approach for data imputation, involving local learning and global approximation in tandem, whereas in the literature only one of them is used. In stage 1, K-means algorithm is used to replace the missing values with cluster centers. Stage 2 refines the resultant approximate values using multilayer perceptron (MLP). MLP is trained on the complete data with the attribute having missing values as the target variable one at a time. The hybrid is tested on 2 benchmark problems each in classification and regression using 10-fold cross validation. In all datasets, some values, which are randomly removed, are treated as missing values. The actual and the predicted values obtained are compared by using Mean Absolute Percentage Error (MAPE). We observe that, the MAPE value is reduced from stage 1 to stage 2, indicating the hybrid approach resulted in better imputation compared to stage 1 alone.