ORIGINAL ARTICLE Particle swarm optimization for feature selection with application in obstructive sleep apnea diagnosis Li-Fei Chen • Chao-Ton Su • Kun-Huang Chen • Pa-Chun Wang Received: 6 March 2011 / Accepted: 3 May 2011 Ó Springer-Verlag London Limited 2011 Abstract Feature selection is a preprocessing step of data mining, in which a subset of relevant features is selected for building models. Searching for an optimal feature subset from a high-dimensional feature space is an NP-complete problem; hence, traditional optimization algorithms are inefficient in solving large-scale feature selection problems. Therefore, meta-heuristic algorithms are extensively adopted to effectively address feature selection problems. In this paper, we propose an analytical approach by integrating particle swarm optimization (PSO) and the 1-NN method. The data sets collected from UCI machine learning databases were used to evaluate the effectiveness of the proposed approach. Implementation results show that the classification accuracy of the proposed approach is significantly better than those of BPNN, LR, SVM, and C4.5. Furthermore, the proposed approach was applied to an actual case on the diagnosis of obstructive sleep apnea (OSA). After implementation, we conclude that our proposed method can help identify important factors and provide a feasible model for diag- nosing medical disease. Keywords Feature selection  Particle swarm optimization  Obstructive sleep apnea  Genetic algorithm 1 Introduction Data mining is the process of automatically discovering useful knowledge from large data sets. This process includes a series of transformation steps, from data pre- processing to the post-processing of mining results. Data preprocessing, in which feature extraction and selection methods play important roles, is perhaps the most time- consuming step in the entire process. In traditional approaches, practical difficulties are encountered in ana- lyzing real data sets. Some of the commonly encountered challenges include the presence of high-dimensional data, heterogeneous and complex data, class imbalance, and class-overlapping data. In addition, the number of features directly affects classification accuracy and processing time, especially in medical science, image processing, and complicated pattern-recognition applications. In handling large amounts of data, feature selection is usually applied to select a set of essential features based on certain criteria. Feature selection is intended to enable the design of a more compact classifier with as little perfor- mance degradation as possible. It reduces the number of features by eliminating irrelevant and redundant attributes, L.-F. Chen (&) Graduate Program of Business Management, Fu-Jen Catholic University, No. 510, Zhongzheng Rd., Xinzhung Dist., New Taipei City 24205, Taiwan, R.O.C e-mail: 075033@mail.fju.edu.tw C.-T. Su  K.-H. Chen Department of Industrial Engineering and Engineering Management, National Tsing Hua University, Hsinchu, Taiwan, R.O.C P.-C. Wang Department of Otolaryngology, Cathay General Hospital, Taipei, Taiwan, R.O.C P.-C. Wang School of Medicine, Fu Jen Catholic University, Taipei, Taiwan, R.O.C P.-C. Wang Department of Public Health, China Medical University, Taichung, Taiwan, R.O.C 123 Neural Comput & Applic DOI 10.1007/s00521-011-0632-4