Proceedings of the Third International Conference on Computing and Wireless Communication Systems, ICCWCS 2019, April 24-25, 2019, Faculty of Sciences, Ibn Tofaïl University -Kénitra- Morocco

Research Article

The Hybrid Machine Learning Model Based on Random Forest Optimized by PSO and ACO for Predicting Heart Disease

Download974 downloads
  • @INPROCEEDINGS{10.4108/eai.24-4-2019.2284088,
        author={Youness  KHOURDIFI and Mohamed  BAHAJ},
        title={The Hybrid Machine Learning Model Based on Random Forest Optimized by PSO and ACO for Predicting Heart Disease},
        proceedings={Proceedings of the Third International Conference on Computing and Wireless Communication Systems, ICCWCS 2019, April 24-25, 2019, Faculty of Sciences, Ibn Tofa\~{n}l University -K\^{e}nitra- Morocco},
        publisher={EAI},
        proceedings_a={ICCWCS},
        year={2019},
        month={5},
        keywords={machine learning; heart disease; random forest; ant colony optimization; particle swarm optimization},
        doi={10.4108/eai.24-4-2019.2284088}
    }
    
  • Youness KHOURDIFI
    Mohamed BAHAJ
    Year: 2019
    The Hybrid Machine Learning Model Based on Random Forest Optimized by PSO and ACO for Predicting Heart Disease
    ICCWCS
    EAI
    DOI: 10.4108/eai.24-4-2019.2284088
Youness KHOURDIFI1,*, Mohamed BAHAJ1
  • 1: Faculty of Sciences and Techniques, Hassan 1st University, Settat, Morocco
*Contact email: ykhourdifi@gmail.com

Abstract

In this paper, we used the hybrid Machine Learning model, for proposed PA-RF, a classification based on Random Forest model, optimized by Particle Swarm Optimization (PSO) associated with Ant Colony Optimization (ACO), and we use Fast Correlation-Based Feature Selection (FCBF) method to filter redundant and irrelevant characteristics, in order to improve the quality of heart disease classification. The proposed mixed approach is applied to the heart disease dataset. The results demonstrate the effectiveness and robustness of the proposed hybrid method in processing various types of data for the classification of heart disease. Therefore, this study examines the different automatic learning algorithms and compares the results using different performance measures, i.e. Accuracy, Precision, Recall, F1-Score, etc. The data set used in this study comes from the UCI's automatic learning repository, entitled "Heart Disease" Data set. We can be concluded that PA-RF has demonstrated efficiency and robustness compared to other classification methods.