Mohammad Ali Afshar Kazemi - Associated Professor Department of Industrial Management, Tehran Central Branch, IAU, Iran
Fereydoon Rahnamay Roodposhti - Professor Department of Management and Economic, Tehran Science and Research Branch, IAU, Iran
Alireza Kheyrkhah - Corresponding Author Master degree, Business Management Tehran, Iran

In this paper, data mining approaches have been presented for identifying the best clustering indicator(s) in Tehran Stock Exchange. For this purpose, seven indicators were selected atdifferent times from the Tehran Stock Exchange for April 2010 to April 201 included return, standard deviation, Beta, P/E, the number of buyers, the number of deals and the volume of transaction, and clustered by three methods (K-means, Fuzzy C-means, and Self-organizedmaps). After clustering, the stocks could be selected from these groups for building a portfolio. Portfolios optimized by Markowitz model to impose the lowest risk to investor for a certainreturn. The best portfolio was selected by Sharp Ratio. Result depicts that Fuzzy C-meansmethod with return and standard deviation present the better portfolio compared to K-means and Self-organized maps (SOM), and creates the most compact cluster compared to others

Portfolio Management, K-means Clustering, Self-organized maps (SOM), Fuzzy Cmeans