Geodesic
A depth-based modification of the k-nearest neighbour method
Kybernetika, Tome 57 (2021) no. 1, pp. 15-37
Cet article a éte moissonné depuis la source Czech Digital Mathematics Library

Voir la notice de l'article

We propose a new nonparametric procedure to solve the problem of classifying objects represented by $d$-dimensional vectors into $K \geq 2$ groups. The newly proposed classifier was inspired by the $k$ nearest neighbour (kNN) method. It is based on the idea of a depth-based distributional neighbourhood and is called $k$ nearest depth neighbours (kNDN) classifier. The kNDN classifier has several desirable properties: in contrast to the classical kNN, it can utilize global properties of the considered distributions (symmetry). In contrast to the maximal depth classifier and related classifiers, it does not have problems with classification when the considered distributions differ in dispersion or have unequal priors. The kNDN classifier is compared to several depth-based classifiers as well as the classical kNN method in a simulation study. According to the average misclassification rates, it is comparable to the best current depth-based classifiers.
We propose a new nonparametric procedure to solve the problem of classifying objects represented by $d$-dimensional vectors into $K \geq 2$ groups. The newly proposed classifier was inspired by the $k$ nearest neighbour (kNN) method. It is based on the idea of a depth-based distributional neighbourhood and is called $k$ nearest depth neighbours (kNDN) classifier. The kNDN classifier has several desirable properties: in contrast to the classical kNN, it can utilize global properties of the considered distributions (symmetry). In contrast to the maximal depth classifier and related classifiers, it does not have problems with classification when the considered distributions differ in dispersion or have unequal priors. The kNDN classifier is compared to several depth-based classifiers as well as the classical kNN method in a simulation study. According to the average misclassification rates, it is comparable to the best current depth-based classifiers.
DOI : 10.14736/kyb-2021-1-0015
Classification : 62G30, 62H30
Keywords: Bayes classifier; data depth; k nearest depth neighbours; nonparametric 
@article{10_14736_kyb_2021_1_0015,
     author = {Venc\'alek, Ond\v{r}ej and Hlubinka, Daniel},
     title = {A depth-based modification of the k-nearest neighbour method},
     journal = {Kybernetika},
     pages = {15--37},
     year = {2021},
     volume = {57},
     number = {1},
     doi = {10.14736/kyb-2021-1-0015},
     mrnumber = {4231854},
     zbl = {07396253},
     language = {en},
     url = {http://geodesic.mathdoc.fr/articles/10.14736/kyb-2021-1-0015/}
}
TY  - JOUR
AU  - Vencálek, Ondřej
AU  - Hlubinka, Daniel
TI  - A depth-based modification of the k-nearest neighbour method
JO  - Kybernetika
PY  - 2021
SP  - 15
EP  - 37
VL  - 57
IS  - 1
UR  - http://geodesic.mathdoc.fr/articles/10.14736/kyb-2021-1-0015/
DO  - 10.14736/kyb-2021-1-0015
LA  - en
ID  - 10_14736_kyb_2021_1_0015
ER  - 
%0 Journal Article
%A Vencálek, Ondřej
%A Hlubinka, Daniel
%T A depth-based modification of the k-nearest neighbour method
%J Kybernetika
%D 2021
%P 15-37
%V 57
%N 1
%U http://geodesic.mathdoc.fr/articles/10.14736/kyb-2021-1-0015/
%R 10.14736/kyb-2021-1-0015
%G en
%F 10_14736_kyb_2021_1_0015
Vencálek, Ondřej; Hlubinka, Daniel. A depth-based modification of the k-nearest neighbour method. Kybernetika, Tome 57 (2021) no. 1, pp. 15-37. doi: 10.14736/kyb-2021-1-0015

[1] Agostinelli, C., Romanazzi, M.: Local depth. J. Statist. Plann. Inference 141 (2011), 817-830. | DOI | MR

[2] Barber, C. B., Dobkin, D. P., Huhdanpaa, H.: The quickhull algorithm for convex hulls. ACM Trans. Math. Software (TOMS) 22 (1996), 4, 469-483. | DOI | MR

[3] Christmann, A., Rousseeuw, P. J.: Measuring overlap in binary regression. Comput. Statist. Data Analysis 37 (2001), 65-75. | DOI | MR

[4] Cox, L. H., Johnson, M. M., Kafadar, K.: Exposition of statistical graphics technology. In: ASA Proc Stat. Comp Section 1982, pp. 55-56. | DOI

[5] Dutta, S., Ghosh, A. K.: On classification based on Lp depth with an adaptive choice of p. Preprint, 2011.

[6] Dutta, S., Ghosh, A. K.: On robust classification using projection depth. Ann. Inst.Statist. Math. 64 (2012), 3, 657-676. | DOI | MR

[7] Dutta, S., Ghosh, A. K., Chaudhuri, P.: Some intriguing properties of Tukey's half-space depth. Bernoulli 17 (2011), 4, 1420-1434. | DOI | MR | Zbl

[8] Fix, E., Hodges, J. L.: Discriminatory Analysis: Nonparametric Discrimination: Consistency Properties. Technical Report 4, Randolph Field, Texas: USAF School of Aviation Medicine, 1951.

[9] Fraiman, R., Liu, R. Y., Meloche, J.: Multivariate density estimation by probing depth. Lecture Notes-Monograph Series 1997, pp. 415-430. | DOI | MR

[10] Ghosh, A. K., Chaudhuri, P.: On maximum depth and related classifiers. Scand. J. Statist. 32 (2005), 327-350. | DOI | MR

[11] Ghosh, A. K., Chaudhuri, P.: On data depth and distribution-free discriminant analysis using separating surfaces. Bernoulli 11 (2005), 1, 1-27. | DOI | MR

[12] Habel, K., Grasman, R., Gramacy, R. B., Mozharovskyi, P., Sterratt, D. C.: Geometry: Mesh Generation and Surface Tessellation. R package version 0.4.5.

[13] Hlubinka, D., Kotík, L., Vencálek, O.: Weighted data depth. Kybernetika 46 (2010), 1, 125-148. | MR

[14] Hubert, M., Veeken, S. van der: Fast and robust classifiers adjusted for skewness. In: COMPSTAT 2010: Proceedings in Computational Statistics: 19th Symposium held in Paris 2010 (Y. Lechevallier and G. Saporta, eds.), Springer, Heidelberg 2010, pp. 1135-1142.

[15] Jörnsten, R.: Clustering and classification based on the $L_1$ data depth. J. Multivar. Anal. 90 (2004), 67-89. | DOI | MR

[16] Kotík, L., Hlubinka, D.: A weighted localization of halfspace depth and its properties. J. Multivar. Anal. 157 (2017), 53-69. | DOI | MR

[17] Kosiorowski, D., Zawadzki, Z.: DepthProc An R Package for Robust. Exploration of Multidimensional Economic Phenomena, 2020.

[18] Lange, T., Mosler, K., Mozharovskyi, P.: Fast nonparametric classification based on data depth. Statist. Papers 55 (2014), 1, 49-69. | DOI | MR

[19] Li, J., Cuesta-Albertos, J. A., Liu, R. Y.: DD-classifier: Nonparametric classification procedure based on DD-plot. J. Amer. Statist. Assoc. 107 (2012), 498, 737-753. | DOI | MR

[20] Liu, R. Y.: On a notion of data depth based on random simplices. Ann. Statist. 18 (1990), 1, 405-414. | DOI | MR

[21] Liu, R. Y., Parelius, J. M., Singh, K.: Multivariate analysis by data depth: Descriptive statistics, graphics and inference (with discussion). Ann. Statist. 27 (1999), 783-858. | DOI | MR

[22] Mardia, K., Kent, J., Bibby, J.: Multivariate Analysis. Academic Press, 1979. | MR

[23] Paindaveine, D., Bever, G. Van: From depth to local depth: a focus on centrality. J. Amer. Statist. Assoc. 105 (2013), 1105-1119. | DOI | MR

[24] Paindaveine, D., Bever, G. Van: Nonparametrically consistent depth-based classifiers. Bernoulli 21 (2015), 1, 62-82. | DOI | MR

[25] Pokotylo, O., Mozharovskyi, P., Dyckerhoff, R.: Depth and depth-based classification with R package ddalpha. J. Statist. Software 91 (2019), 5, 1-46. | DOI

[26] Serfling, R.: Depth functions in nonparametric multivariate inference. In: Data Depth: Robust Multivariate Analysis, Computational Geometry and Applications (R. Y. Liu, R. Serfling, and D. L. Souvaine, eds.), American Mathematical Society, DIMACS Series in Discrete Mathematics and Theoretical Computer Science 7, New York 2006, pp. 1-16. | MR

[27] Vencalek, O.: $k$-Depth-nearest neighbour method and its performance on skew-normal distributons. Acta Univ. Palacki Olomouc., Fac. Rer. Nat., Mathematica 52 (2013), 2, pp. 121-129. | MR

[28] Yeh, I. C., Yang, K. J., Ting, T. M.: Knowledge discovery on RFM model using Bernoulli sequence. Expert Systems Appl. 36 (2009), 5866-5871. | DOI

[29] Zakai, A., Ritov, Y.: Consistency and localizability. J. Machine Learning Res. 10 (2009), 827-856. | MR

[30] Zuo, Y., Serfling, R.: General notion of statistical depth function. Ann. Statist. 28 (2000), 461-482. | DOI | MR

[31] Zuo, Y., Serfling, R.: Structural properties and convergence results for contours of sample statistical depth functions. Ann. Statist. 28 (2000) 2, 483-499. | DOI | MR

Cité par Sources :