A hybrid model for classification of medical data set based on factor analysis and extreme learning machine: FA + ELM

Abstract

Data mining techniques such as classification, clustering, and prediction are used extensively for medical diagnosis in epidemiological fields. A hybrid model based on Factor Analysis (FA) and Extreme Learning Machine (ELM) was proposed in this study for diagnosing breast cancer, Lymphography, and erythemato-squamous diseases. The proposed hybrid model consists of two stages. Firstly, FA was used for preprocessing the medical dataset, and then, the factors obtained using FA were used as input features for the ELM model. Dermatology, Lymphography, and Wisconsin Breast Cancer real datasets obtained from the UCI machine learning database were used to test the proposed model. An average success rate of 96.39 % and 96.94 % was obtained after classifying the dermatology dataset with ELM and FA + ELM models. While the success rate obtained by classifying the lymphography data set using ELM is 84.50 %, the result obtained with FA + ELM is 85.10 %. The success rates of 97.10 % and 97.25 % are achieved respectively for Wisconsin Breast Cancer (WBC) using ELM and FA + ELM. As a result, it was observed that preprocessing of the data increased the average classification success in three different medical datasets used for the classification problem. It is considered that the proposed hybrid model will be helpful for the decision-making stage in medical diagnosis systems. © 2022 Elsevier Ltd