DOI QR코드

DOI QR Code

수정된 적응 최근접 방법을 활용한 판별분류방법에 대한 연구

On the Use of Modified Adaptive Nearest Neighbors for Classification

  • 투고 : 20100800
  • 심사 : 20100900
  • 발행 : 2010.12.31

초록

비모수적 판별분류방법인 k-Nearest Neighbors Classification(KNNC) 방법은 널리 사용되고 있지만 고정된 이웃의 개수를 사용하며 또한 집단변수의 정보를 활용하지 않음으로서 자료의 국소적 특징을 반영하지 못하는 단점이 있다. Adaptive Nearest Neighbors Classification(ANNC) 방법과 Modified k-Nearest Neighbors Classification(MKNNC) 방법은 각각 이러한 단점들을 보완하기 위해 제안된 방법이다. 본 연구에서는 ANNC 방법과 MKNNC 방법의 장점을 결합한 Modified Adaptive Nearest Neighbors Classification(MANNC) 방법을 제안하였다. 나아가, 제안된 방법의 활용 가능성을 살펴보고자 실제자료에 대한 분석과 모의실험을 통해 기존의 방법들과 비교하였다.

Even though the k-Nearest Neighbors Classification(KNNC) is one of the popular non-parametric classification methods, it does not consider the local features and class information for each observation. In order to overcome such limitations, several methods have been developed such as Adaptive Nearest Neighbors Classification(ANNC) and Modified k-Nearest Neighbors Classification(MKNNC). In this paper, we propose the Modified Adaptive Nearest Neighbors Classification(MANNC) that employs the advantages of both the ANNC and MKNNC. Through a real data analysis and a simulation study, we show that the proposed MANNC outperforms other methods in terms of classification accuracy.

키워드

참고문헌

  1. 전명식, 최인경 (2009). Adaptive nearest neighbors를 활용한 판별분류방법, <응용통계연구>, 22, 479-488. https://doi.org/10.5351/KJAS.2009.22.3.479
  2. 정윤경, 백장선 (2007). 고차원(유전자 발현)자료에 대한 군집 타당성 분석 기법의 성능비교, <응용통계연구>, 20, 167-181. https://doi.org/10.5351/KJAS.2007.20.1.167
  3. Friedman, J. (1994). Flexible Metric Nearest Neighbor Classification, Technical report, Standford University.
  4. Hastie, T. and Tibshirani, R. (1996). Discriminant adaptive nearest neighbor classification, IEEE Transactions on Pattern Analysis and Machine Intelligence, 18, 607-616. https://doi.org/10.1109/34.506411
  5. Jhun, M., Jeong, H. C. and Koo, J. Y. (2007). On the use of adaptive nearest neighbors for missing value imputation, Communications in Statistics: Simulation and Computation, 36, 1275-1286. https://doi.org/10.1080/03610910701569069
  6. Parvin, H., Alizadeh, H. and Minaei-Bidgoli, B. (2008). MKNN: Modified k-nearest neighbor, Proceedings of the World Congress on Engineering and Computer Science, 22-24.
  7. Song, Y., Huang, J., Zhou, D., Zha, H. and Giles, C. L. (2007). IKNN: Informative k-nearest neighbor pattern classification, PKDD 2007, 248-264.

피인용 문헌

  1. On the Use of Sequential Adaptive Nearest Neighbors for Missing Value Imputation vol.24, pp.6, 2011, https://doi.org/10.5351/KJAS.2011.24.6.1249