Open Conference Systems, 50th Scientific meeting of the Italian Statistical Society

Font Size: 
Distance based Depth-Depth classifier for directional data
Giuseppe Pandolfo, Giovanni C. Porzio

Last modified: 2018-06-04

Abstract


The DD-classifier, which has been extended to the classification of directional objects, is here investigated in the case of some new distance-based directional depths. The DD-classifier is a non-parametric techniques based on the depth vs.  depth (DD) plot. Its main advantages concern the flexibility and the independence from specific distribution parameters. The new depth functions adopted here allow using them for high-dimensional directional data sets.

References


1. Ackermann, H.: A note on circular nonparametrical classification. Biometrical J 5, 577–587(1997)

2. Batschelet, E.: Circular statistics in biology. Academic Press, London, (1981)

3. Liu, R.Y., Singh, K.: Ordering directional data. Concepts of data depth on circles and spheres.Ann Stat 20, 1468–1484 (1992)

4. Liu, R.Y., Singh, K.: Multivariate analysis by data depth: descriptive statistics, graphics andinference. Ann Stat 27, 783–1117 (1999)

5. Mardia, K.V., Jupp, E.P.: Statistics of directional data. Academic Press, London (1972)

6. Pandolfo, G., Paindaveine, D., Porzio, G.C.: Distance-based depths for directionaldata. Working Papers ECARES 2017 ô€€€ 35 (2017) Available athttps://ideas.repec.org/p/eca/wpaper/2013-258549.html

7. Stoller, D.S.: Univariate two-population distribution-free discrimination. J Am Statist Assoc49, 770-–777 (1954)


Full Text: PDF