臺灣博碩士論文加值系統

最近幾年，模糊決策樹(fuzzy decision tree)已被廣泛應用在萃取資料規則及分類上，透過此類的演算法，找出可能有問題的資料(如盜刷信用卡、醫療費用浮報等)及預測新事物的類別。而很多學者都在致力於建構最佳化的模糊決策樹演算法，以提高預測的準確性。這些研究主要都是在探討決策樹的建置過程，包括了在切割連續屬性時歸屬函數的決定，或是建構模糊決策樹時所採用的模糊熵(fuzzy entropy)演算法等，而卻很少在探討建置模糊決策樹後，在推論過程所必經的步驟─解模糊化(defuzzification)。因此本研究著重在提出新的解模糊化方法，運用了加權模糊推論規則的概念在解模糊化上，所建構出的決策樹稱為加權模糊決策樹(weighted fuzzy decision tree)。
此加權模糊決策樹結合了加權模糊推論規則(weighted fuzzy production rule)及模糊貝氏推論法(fuzzy Bayesian inference)。在解模糊化的過程中主要是運用了加權模糊推論規則的概念來完成，而其中所需的參數─權重，則是利用模糊貝氏推論法來取得。此加權模糊決策樹將落實資料探勘的目的，從一堆資料中，找出規則，進而預測及分類新資料，進而產生新知識。在驗證方面，採用標準測試資料實證此方法的正確性，當樹建構成非完全決策樹時，我們的方法較其他解模糊化方法(x-x-+及KNN)分類的正確性來的佳；當樹建構成完全決策樹時，也有不錯的成效。

In recent years, fuzzy decision tree had been widely used to extracting classification knowledge from a set of feature-based data. And many researchers are engaged in the more efficient and optimal algorithms to construct fuzzy decision trees. However, very few papers discuss the process of defuzzification in fuzzy decision tree. Therefore, we propose a new method that emphasizes on the defuzzification process. The tree build by our method is called weighted fuzzy decision tree. It uses the concept of weighted fuzzy production rule(WFPR) in defuzzification process and the concept of fuzzy Bayesian inference(FBI) method to find the parameters needed in the inference process of WFPR. To verify the accuracy of our method for classification, standard benchmark datasets are used. When the tree is build as non-perfect decision tree, our proposed method has higher accuracy for classification than other defuzzification methods; when the tree is perfect decision tree, our method is also acceptable.

1.INTRODUCTION1
2.FUZZY DECISION TREE4
2.1 CONSTRUCTION OF FUZZY DECISION TREE6
2.2 DEFUZZIFICATION OF FUZZY DECISION TREE7
2.2.1 x-x-+ Method7
2.2.2 KNN Method8
3.WEIGHTED FUZZY PRODUCTION RULE (WFPR)9
3.1 PARAMETERS IN WFPR10
3.2 DEFINITION OF WFPR10
3.3 FUZZY REASONING IN WFPR12
4.FUZZY BAYESIAN INFERENCE (FBI)14
4.1 DISCRETE VALUES OR FUZZY VALUES15
4.2 CONTINUOUS VALUES15
5.RESEARCH DESIGN17
5.1 RESEARCH FRAMEWORK17
5.2 RESEARCH METHOD18
6.EXPERIMENTS20
6.1 THE ECHO DATA20
6.1.1 Non-Perfect Weighted Fuzzy Decision Tree21
6.1.2 Perfect Weighted Fuzzy Decision Tree23
6.2 THE BREAST CANCER DATA25
6.2.1 Non-Perfect Weighted Fuzzy Decision Tree25
6.2.2 Perfect Weighted Fuzzy Decision Tree27
6.3 THE H-HEART DATA29
6.4 DISCUSSION30
7.CONCLUSION31
7.1 RESEARCH CONCLUSION31
7.2 RESEARCH LIMITATIONS32
7.3 FUTURE WORKS33
REFERENCES34
Figure 1: Vague cut point4
Figure 2: Fuzzy decision tree6
Figure 3: Defuzzification by x-x-+ method7
Figure 4: Conceptual framework for combining fuzzy Bayesian inference and weighted fuzzy production rule17
Figure 5: Non-perfect weighted fuzzy decision tree of the echo data22
Figure 6: Perfect weighted fuzzy decision tree of the echo data24
Figure 7: Non-perfect weighted fuzzy decision tree of the breast cancer data26
Table 1: The characters of the four types of fuzzy decision tree5
Table 2: Parameters in WFPR10
Table 3: The classification accuracy of the echo data of the non-perfect weighted fuzzy decision tree23
Table 4: The classification accuracy of the echo data of the perfect weighted fuzzy decision tree25
Table 5: The classification accuracy of the breast cancer data of the non-perfect weighted fuzzy decision tree27
Table 6: The classification accuracy of the breast cancer data of the perfect weighted fuzzy decision tree29
Table 7: The classification accuracy of the H-heart data of the non-perfect weighted fuzzy decision tree30

[1]Chang R.L.P. and Pavlidis T. (1977), “Fuzzy Decision Tree Algorithms.”, IEEE Trans. on systems, Man and Cybernetics, Vol.7, No.1, p.28-35.
[2]Chen S.M. (1988), “A new approach to handling fuzzy decision making problems.”, IEEE Trans. Systems Man Cybernet, SMC-18, p.1012-1016.
[3]Chen S.M., Ke J.S. and Chang J.F. (1990), “Knowledge representation using fuzzy Petri nets.”, IEEE Trans. Knowledge Data Engrg. 2, p.311-319.
[4]Hudson D.L. and Cohen M.E. (1991), “The rule of approximate reasoning in medical expert system.”, Fuzzy Expert System, Vol.11(CRC, Florida), p.165-179.
[5]Ittner A., Zeidler J., Rossius R., Dilger W. and Schlosser M. (1997), “Feature space partitioning by non-linear and fuzzy decision trees”, Proc. of the 7th IFSA World Congress(Prague, Czech), p.394-398.
[6]Janikow C.Z. (1998), “Fuzzy Decision Trees: Issues and Methods.”, IEEE Trans. on Systems, Man and Cybernetics Part B, Vol.28, No.1, p.1-14.
[7]Lan C.H. and Chan C.L.(2000), “A Data Mining Technique Combining Fuzzy Sets Theory and Bayesian Classfier”, The Eleventh National Conference on MIS, MIS Department in Sun Yat-sen University.
[8]Lee K.M., Lee K.M., Lee J.H. and Hyung L.K. (1999), “A fuzzy decision tree induction method for fuzzy data.”, Fuzzy Systems Conference Proceedings, FUZZ-IEEE ''99, 1999 IEEE International, Vol.1, p.16 —21.
[9]Leung K.S. and Lam W. (1988), “Fuzzy concepts in expert systems.”, IEEE Comput., p.43-56.
[10]Quinlan J.R. (1979), “Discovering rules by induction from large collections of examples.”, Expert Systems in the Micro Electronics Age, Edinburgh University Press.
[11]Quinlan J.R. (1986), “Induction of decision trees.”, Machine Learning, Vol.1, p.81-106.
[12]Quinlan J.R. (1993), “C4.5: Programs for Machine Learning.”, Morgan Kaufmann Publishers.
[13]Tseng H.C. and Teo D.W. (1994), “Medical expert system with elastic fuzzy logic.”, 3rd IEEE Internet. Conf. On Fuzzy Systems 3, p.2067-2071.
[14]Umano M., Okamoto H., Hatono I., Tamura H., Kawachi F., Umedzu S. and Kinoshita J. (1994), “Fuzzy decision trees by fuzzy ID3 algorithm and its application to diagnosis systems.”, In Proc. Of 3rd IEEE International Conference on Fuzzy Systems, p.2113-2118.
[15]Webber R. (1992), “Fuzzy-ID3: a class of methods for automatic knowledge acquisition, Proc. of 2nd Int. Conf. on Fuzzy Logic and Neural Networks(Iizuka), p.265-268.
[16]Yang C.C. (1995), “Fuzzy Bayesian analysis with continuous-valued evidence.”, Proceedings of the IEEE International Conference on Systems, Man and Cybernetics, v1, p.441-446
[17]Yang C.C. (1997), “Fuzzy Bayesian inference.”, Proceedings of the IEEE International Conference on Systems, Man and Cybernetics v3, p.2707-2712.
[18]Yeung D.S. and Tsang E.C.C. (1992), “Fuzzy production rules evaluation and fuzzy Petri nets.”, in: Adrances in Computer Cybernetics and Information Engineering, 1993 collection of selected papers presented at the 6th Internat. Conf. On Systems Research, Informatics and Cybernetics, Baden-Baden, Germany, p.137-142.
[19]Yeung D.S. and Tsang E.C.C. (1995), “A weighted fuzzy production rule evaluation method.”, Fuzzy Systems, International Joint Conference of the Fourth IEEE International Conference on Fuzzy Systems and The Second International Fuzzy Engineering Symposium., Proceedings of 1995 IEEE International Conference, Vol.2, p.461-468.
[20]Yeung D.S. and Tsang E.C.C. (1997), “Weighted fuzzy production rules.”, Fuzzy Sets and Systems, Vol.88, p.299-313.
[21]Yuan Y. and Shaw M.J. (1995), “Induction of fuzzy decision trees.”, Fuzzy Sets and Systems, Vol.69, p.125-139.
[22]Zeidler J., Schlosser M., Ittner A. and Posthoff C. (1996), “Fuzzy decision trees and numerical attributes.”, Proceedings of the 5th IEEE International Conference, Vol.2, p.985-990.
[23]Zeidler J. and Schlosser M. (1996), “Continuous-valued attributes in fuzzy decision trees.”, Proc. of the 6th Int. Conf. on Information Processing and Management of Uncertainty in knowledge-Based Systems(Granada, Spain), p.395-400.
[24]Zhung Y.M. (1992), “The Fuzzy Decision Tree Induction”, A Thesis Submitted to Institute of Information Sun Yat-sen University in Partial Fulfillment of the Requirements for the Degree of Master in Information Management.