Title: Rough set theory-based feature selection and FGA-NN classifier for medical data classification

Authors: Vijayalakshmi Balasubramanian; Sugumar Rajendran

Addresses: Department of Computer Science, Bharathiar University, Coimbatore, India ' Department of Computer Science and Engineering, Velammal Institute of Technology, Chennai, India

Abstract: The prediction of heart disease is a difficult task, which needs much experience and knowledge. In order to reduce the risk of heart disease prediction, in this paper we proposed a rough set theory-based feature selection and FGA-NN classifier. The overall process of the proposed system consists of two main steps, such as: 1) feature reduction; 2) heart disease prediction. At first, the kernel fuzzy c-means clustering with roughest theory (KFCMRS) algorithm is applied to the high dimensional data to reduce the dimension of the attribute. After that, the medical data classification is done through FGA-NN classifier. To improve the prediction performance, hybridisation of firefly and genetic algorithm (FGA) is utilised with NN for weight optimisation. At last, the experimentation is performed by means of Cleveland, Hungarian, and Switzerland datasets. The experimentation result proves that the FGA-NN classifier outperformed the existing approach by attaining the accuracy of 83%.

Keywords: heart disease; FGA-NN; KFCMRS; scaled conjugate gradient; SCG; prediction; feature reduction; optimisation.

DOI: 10.1504/IJBIDM.2019.098838

International Journal of Business Intelligence and Data Mining, 2019 Vol.14 No.3, pp.322 - 358

Received: 18 Nov 2016
Accepted: 03 Feb 2017
Published online: 04 Apr 2019 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article