A modified particle swarm optimization algorithm used for feature selection of UCI biomedical data sets

Publication Type:
Conference Proceeding
60th International Scientific Conference on Information Technology and Management Science of Riga Technical University, ITMS 2019 - Proceedings, 2019
Issue Date:
Full metadata record
© 2019 IEEE. In biomedical and health care applications, classification examination is extensively used to help diagnose health problems, decision making and enhance standards of patient care. Feature selection is a significant data pre-processing method in classification problems. Training of the data is achieved by using a subset dataset from the UCI biomedical database. If the training dataset comprises inappropriate features, classification analysis resulted in inaccurate and incomprehensible results. In data mining, feature subset selection is data pre-processing phase that is of enormous importance. In this paper, for selecting a minimum number of features K-Nearest Neighbour (KNN) classifier is presented with a modified particle swarm optimization (MPSO) to obtain good classification precision. The proposed method is applied to three UCI medical data sets and is compared with other feature selection approach available in the literature. Results demonstrate that the feature subset recognized by the presented MPSO with KNN neighbor classifiers give better results and accuracy as compared to the other techniques.
Please use this identifier to cite or link to this item: