... visualizing, financial benchmarking I RESULTS We reviewed some approaches to analyze financial data as so as considered backgrounds for using SOM in benchmarking companies We proposed a model for... data preprocessing, training the network, visualizing, producing financial benchmarking results Some options for developing the model will be presented An application that supports training and... and removed outliers Then they were normalized After these processes, the data were trained with SOM The quality of the training process was estimated in both measures: average quantization error
1 S D ng SOM (Self-Organizing Map) Xây D ng Mô Hình ánh Giá Tài Chính Doanh Nghi p H ng d n: PGS.TS Lê Hoài B c; Th c hi n: Nguy n Kh c Thìn Tóm t t—M c tiêu c a nghiên c u này là s xây d ng m t mô hình đánh giá tài chính doanh nghi p trong ngành cung c p d ch v vi n thông s d ng b n đ t t ch c. Mô hình s bao g m các quá trình: l a ch n công ty, thu th p d li u, phát hi n và lo i b bi t l , chu n hoá d li u, hu n luy n và cho k t qu đánh giá tài chính. Gi i thi u các tu bi n khi tri n khai mô hình, cài đ t ng d ng h tr các tác v c a mô hình. T khoá—Khai phá d li u, b n đ t t ch c, bi t l , chu n hoá d li u, gom c m, đánh giá tài chính. I. K T QU Nghiên c u đã xem xét m t s cách ti p c n kh d đ phân tích d li u tài chính c ng nh xem xét c s c a vi c s d ng SOM trong đánh giá tài chính doanh nghi p. Nghiên c u đã đ xu t mô hình đánh giá tài chính các doanh nghi p v i nhi u tu bi n. Nghiên c u đã l a ch n và thu th p d li u tài chính liên quan cho nghiên c u c a 101 công ty hàng đ u toàn c u trong giai đo n 2000-2004 trong l nh v c cung c p d ch v vi n thông. D li u thu th p đ c x lý bi t l , chu n hoá tr c khi đ a vào hu n luy n. Ch t l ng hu n luy n đ c đánh giá theo l i l ng t hoá và l i topo b n đ . B n đ thu đ c qua hu n luy n đ c gom c m. Tr c quan hoá b n đ ta thu đ c các đánh giá s b các c m, xu h ng ho t đ ng c a các doanh nghi p. Nghiên c u đã cài đ t ng d ng h tr quá trình hu n luy n d li u, đánh giá ch t l ng hu n luy n, gom c m, gán nhãn và mã hoá màu b n đ SOM thu đ c qua hu n luy n.... Các k t qu vào ra c a ng d ng đ u có th đ c s d ng tr c ti p t các gói ph n m m hu n luy n và tr c quan SOM khác nh SOMPAK và SOM Toolbox for Matlab (c hai ng d ng này đ u do đ i h c Helsinki, Ph n Lan phát tri n). Tri n khai các mô hình đánh giá tài chính khác c ng nh áp d ng các tu bi n khác vào mô hình này đ có các k t qu đ i ch ng TÀI LI U THAM KH O [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] II. H NG PHÁT TRI N i chi u k t qu đánh giá t mô hình v i đánh giá trong báo cáo tài chính hàng n m đ đánh giá đ chính xác c a mô hình. Ngoài ra, mô hình c n nh n đ c s đánh giá c a các chuyên gia tài chính đ c i thi n đ tin c y Xây d ng mô hình đánh giá cho th c ti n Vi t Nam Nguy n Minh Ki u (2005), Phân tích tài chính, Ch ng trình Gi ng d y Kinh t Fulbright, Niên khoá 2005-2006. Charnes, A., W. W. Cooper and E. L. Rhodes (1978), “Measuring the efficiency of decision making units,” European Journal of Operational Research 2(6): 429- 444. Eklund, T., B. Back, H. Vanharanta and A. Visa (2002), “Assessing the Feasibility of Self-Organizing Maps for Data Mining Financial Information,” The Xth European Conference on Information Systems, Gdansk, Poland, June 6-8: 528-537. Eklund, T., B. Back, H. Vanharanta and A. Visa (2004), “Financial Benchmarking Tools in Finnish Companies – A State of the Art Survey,” Turku Centre for Computer Science, Finland. Eklund, T. (2004), “The Self-Organizing Map in Financial Benchmarking,” Doctoral Dissertation, Turku Centre for Computer Science, Finland. Eklund, T. (2005), “Financial Ratio Analysis, Financial Benchmarking, and Self-Organizing Maps - A Current Research Example,” Åbo Akademi University, Finland. Fayyad, U., G. Piatetsky-Shapiro and P. Smyth (1996), “The KDD Process for Extracting Useful Knowledge from Volumes of Data,” Communications of the ACM 39(11): 27-34. Kaplan, R. S. and D. P. Norton (1992), “The Balanced Scorecard Measures that Drive Performance,” Harvard Business Review 70(1): 71-80. Karlsson, J. (2001), “Benchmarking Telecommunication Companies Using Self-Organising Maps,” Master Thesis in Information Systems, Åbo Akademi University, Finland. Kohonen, T. (2001), Self-Organizing Maps, Springer-Verlag, Berlin. Marghescu, D. and M.J. Rajanen (2005), “Assessing the use of the SOM technique in data mining,” Proc. of the 23rd IASTED Intl. MultiConf., Databases and Applications, Innsbruck, Austria. Oja, E., S. Kaski and T. Kohonen (2002), “Bibliography of SelfOrganizing Map (SOM) Papers: 1998-2001 Addendum,” Neural Computing Surveys 3, Helsinki University of Technology, Finland. Pei, J. (2003), Data Mining, CSE 626, State University of New York at Buffalo. Spotfire Inc. (2003), Spotfire DecisionSite Statistics 7.2. Vesanto, J., J. Himberg, E. Alhoniemi, J. Parhankangas (2000), “SOM toolbox for Matlab 5,” Technical report A57, Helsinki University of Technology, Finland. Vesanto, J. and E. Alhoniemi (2000), “Clustering of the Self-Organizing Map,” IEEE Transactions on Neural Networks 11(3): 586-600. 2 Financial Benchmarking International Telecom Services Companies Using Self-Organizing Maps Supervisor: Le Hoai Bac; Executor: Nguyen Khac Thin Abstract— This research aims to build a model for financial benchmarking in the international telecommunication services companies using self-organizing maps. The model includes the following processes: choosing the companies, collecting the data, data preprocessing, training the network, visualizing, producing financial benchmarking results. Some options for developing the model will be presented. An application that supports training and visualizing the map will be implemented. Index Terms—Data mining, self-organizing map, outlier, data normalization, clustering, visualizing, financial benchmarking. I. RESULTS We reviewed some approaches to analyze financial data as so as considered backgrounds for using SOM in benchmarking companies. We proposed a model for financial benchmarking companies with many options. We collected seven financial ratios for 101 telecom services companies in the period 2000-2004. The data were detected and removed outliers. Then they were normalized. After these processes, the data were trained with SOM. The quality of the training process was estimated in both measures: average quantization error and topographic error. The map we got after a good training was cluster using the hierarchical approach. By visualizing the map in some views, we can estimate the financial performance of the companies in the clusters. We also implemented an application that supports training, estimating the quality of the training process, clustering, coloring, labeling and visualizing the map. Inputs and outputs for the application can be used by other SOM packages. II. FUTURE RESEARCH Results from the model should be compared with the annual reports of the companies. In addition, the model should be estimated by financial experts. Developing the model for Vietnamese companies. Comparing the model with other models as so as adding new options for the processes of the model Manuscript received October 1, 2006. Lê Hoài B c is with Faculty of Information Technology, Hochiminh University of Natural Sciences, Hochiminh City, Vietnam (e-mail: baclh@ fit.hcmuns.edu.vn). Nguy n Kh c Thìn, is studying at University of Information Technology, Hochiminh City, Vietnam (phone: 84903-689-054; e-mail: nkthin@yahoo.com).