A Minimum Spanning Tree Algorithm Based on Fuzzy Weighted Distance
The traditional minimum spanning tree clustering algorithm uses a simple Euclidean distance metric method to calculate the distance between two entities. For the processing of noise data, the similarity can’t be well described. In this regard, first of all, we integrate fuzzy set theory to improve, and propose a method with indeterminacy fuzzy distance measurement. In the distance metric method, fuzzy set theory is introduced to measure the differences between two entities. Moreover, the attributes are fuzzy weighted on this basis, which overcomes the shortcomings of the simple Euclidean distance measurement method. So, it not only has a good tolerance for data noise to solve the misclassification of noise information in the actual data, but also takes into account the difference of distinguishability contribution degree of attributes in classification. Thus, the accuracy of clustering is improved, and it also has significance in practical project application. Then, the proposed distance metric is applied into the traditional MST clustering algorithm. Compared with the traditional MST clustering algorithm and other classical clustering algorithms, the results show that the MST algorithm based on the new distance metric is more effective.
KeywordsClustering algorithm Minimum spanning tree algorithm Fuzzy set Membership degree Distance metric
We would like to thank the anonymous reviewers for their valuable comments and suggestions. This work is supported by The State Key Research Development Program of China under Grant 2016YFC0801403, Shandong Provincial Natural Science Foundation of China under Grant ZR2018MF009 and ZR2015FM013, the Special Funds of Taishan Scholars Construction Project, and Leading Talent Project of Shandong University of Science and Technology.
- 1.Xin, C.: Research on clustering analysis method based on minimum spanning tree. A master’s degree thesis. Chongqing University, Chongqing (2013)Google Scholar
- 2.Tang, F.: Data Structure Tutorial, pp. 158–276. Beihang University Press, Beijing (2005)Google Scholar
- 3.Barna, S., Pabitra, M.: Dynamic algorithm for graph clustering using minimum cut tree. In: Proceeding of Sixth IEEE International Conference on Data Mining (2006)Google Scholar
- 5.Caetano Jr., T., Traina, A.J.M., Faloutsos, C.: Feature selection using fractal dimension-ten years later. J. Inf. Data Manag. 1(1), 17–20 (2010)Google Scholar
- 7.Moore, A.W.: An Introductory Tutorial on KD-Trees. University of Cambridge, UK (1991)Google Scholar
- 10.Smarandache, F.: A unifying field in logics. Neutrosophy: Neutrosphic probability, set and logic. American Research Press, Rehoboth, D E (1999)Google Scholar
- 12.Zhang, D.: Research on Motor Fault Diagnosis Method Based on Rough Set Theory. Bohai University, Jinzhou (2015)Google Scholar
- 15.Wang, H.B., Smarandache, F., Zhang, Y.Q., et al.: Single valued neutrosophic sets. Multispace Multructure 4(10), 410–413 (2010)Google Scholar
- 19.Yan, W., Wu, W.: Data structure. Tsinghua University Press, Beijing (1992)Google Scholar
- 22.Shi, Y., Shi, W., Jin, F.: Entropy and its application in the study of uncertainty in spatial data. Comput. Eng. 31 (24), 36–37 (2005)Google Scholar