Authors : P. Penchala Prasad, F. Sagayaraj Francis and S. Zahoor-Ul-Huq
Abstract: Medical data classification is the challenging task due to noisy data or missing data are present in the dataset. The feature selection techniques play the important part in the classification process. The more relevant features help to provide the efficient classification of medical data which is essential for the disease detection. In this research, Recursive Feature Elimination with the Gravitational Search Algorithm (RFE-GSA) is proposed for efficient classification of the data. The Recursive Feature Elimination (RFE) method helps to remove the irrelevant features from the medical data and rank them in order of importance that helps to reduce the computation cost of the proposed method. The ranked features from the RFE are given as input to the GSA which select the feature for the classification. The GSA is fast convergence and that helps to find the relevant features in the data. The features selected from the RFE-GSA is provided as input to the Radial Basis Function (RBF) for the classification. The performance of the RFE-GSA method is high compared to the other existing method. The proposed RFE-GSA method has the accuracy of the 98.24% in the breast cancer dataset in UCI dataset and the state-of-art method has achieved the accuracy of 96.87%.
P. Penchala Prasad, F. Sagayaraj Francis and S. Zahoor-Ul-Huq, 2019. Recursive Feature Elimination and Gravitational Search Algorithm for Classification of Medical Data. Journal of Engineering and Applied Sciences, 14: 8828-8834.