Robust clustering of trade data with thinning processes
We address clustering issues in presence of densely populated data points, exhibiting linear structures with high degree of overlapping. To avoid the disturbing effects of high dense areas, we retain (and then cluster) a sample of data with a process preserving the general structure of the data. The problem is approached as a spatial point process. An intensity function and a thinning process to select the sub-sample for the clustering are derived for the analysis of the EU trade data.
PERROTTA Domenico;
CERIOLI Andrea;
TORTI Francesca;
2012-08-08
Pavia University Press - Editoria Scientifica
JRC67736
978-88-96764-22-0,
Additional supporting files
| File name | Description | File type | |
|---|
