Omission Distance

Inertia · Post by **Inertia** » Wed Feb 21, 2007 1:14 pm

Hello,

could anyone please explain the omission distance?
What is the exact definition?
Is there any relevant literature?

Thanks,

Olli

bubach · Post by **bubach** » Mon Mar 05, 2007 1:59 pm

Hello Oliver,

the article from Tenenhaus et al 2005 in Computational Statistics is very helpful. On page 174f, Tenenhaus describe the Blindfolding approach.

To your question:
The data are divided into several groups (=omission-distance), e.g. a, b, c. First all values for a are deleted and estimated with the remaining values:

MV1 MV2 MV3
a c
b e
c ...

Then the same with values in group b and c.

Hope this short explanation is helpful.

Michael

Inertia · Post by **Inertia** » Wed Mar 07, 2007 9:52 am

Hello Michael,

thanks a lot for your helpful explanation.

Oliver