An Enhanced Multi Granulation Rough Set Feature Selection algorithm using Dynamic Sampling
The indispensable attributes are selected from a dataset using feature selection, an important preprocessing step. In this paper, an enhanced feature selection algorithm based on Multi-granulation Rough set is developed and evaluated on multiple dataset from healthcare domain. The proposed algorithm uses Newton CG method for sampling. The features are selected based on finding the minimal reduct and forming the feature subset for further processing. Performance evaluation of the proposed algorithm is done using a comparison of existing three rough set based feature selection algorithms. The evaluation parameters used were classification accuracy, confusion matrix and mean absolute error. Based on the experiments results, the proposed Multi Granulation Rough Set Feature Selection (MGRSFS) algorithm is efficient to handle large data set and is feasible with multiclass dataset.