비유사성 측정을 위한 준거리 연산

다국어 초록 (Multilingual Abstract)

Various data analysis, such as classification, clustering, and recommendation, begins by measuring the dissimilarity or distance of data items. The distance relation of the data set can be represented by a distance matrix indicating the distance of the items, but it is difficult to get the distance matrix of some data set. The quasi distance condition satisfies the triangular inequality and is the minimum condition of metric. We propose a matrix operation min_sum that generates a quasi-distance relation in a given data set and generated the theorem about the relation between this operation and the condition for general distance metric. Then we prove that this operation transforms the potential distance matrix intuitively created from the data set into a matrix satisfying the quasi-distance condition. It is confirmed that it is effective as a distance measure as a result of applying voting tendency clustering using 1-year actual voting data of lawmakers.

참고문헌 (Reference)

1 Rui Xu, "Survey of Clustering Algorithms" 16 (16): 645-678, 2005

2 Segaran, "Programming Collective Intelligence: Building Smart Web 2.0, Applications" O'Reilly Media 14-, 2007

3 J. C. Gower, "Metric and Euclidean Properties of Dissimilarity Coefficients" 3 (3): 5-48, 1986

4 Drew Conway, "Machine Learning for Hackers" O'Reilly Media 241-, 2012

5 Charu C. Aggarwal, "Data Clustering Algorithms and Applications" CRC Press 7-, 2014

6 Guojun Gan, "Data Clustering : Theory, Algorithms, and Applications, Similarity and Dissimilarity Measures, ASA-SIAM Series on Statistics and Applied Probability" 67-106, 2007

7 Brian S. Everitt, "Cluster Analysis, 5th Ed." John Wiley & Sons, Ltd 49-, 2011

8 A. D. Gordon, "Classification : Methods for the Exploratory Analysis of Multivariate Data, Measuring Dissimilarity, Chap 2" Chapman and Hall 13-32, 1981

9 J. C. Gower, "A General Coefficient of Similarity and Some of Its Properties" 27 (27): 857-871, 1971