[Openspace] K-nearest neighbor matrix

[Patricia Abelairas]

Hello,

I have two question regarding the k-nearest neighbor matrix. 
I would like to know if there is any way to know the best number of
neighbors recommended to use. I have 78 observations in my study. 
The locations of these 78 observations are not homogeneus at all, some of
them are near and some of them are far appart. I have read that this kind
of matrix (KKN) is the best for this type of data, is it?

Thank you very much 
Patricia Abelairas Etxebarria
Departamento de Economía Aplicada V
Facultad de Ciencias Económicas y Empresariales
Universidad del País Vasco
Avda. Lehendakari Aguirre, 83
48015 Bilbao
SPAIN

Tel: +34 94 601 70 26
Fax: +34 94 601 70 87
e-mail: ebbabetp at bs.ehu.es