I at moment implementing kNN in c++. I know how the algorithm works, but was wondering whether there was a smart method for computing the distance between a new observation and and its k nearest neighbours?...
The only way i see that possible, is to test all the point against the observation and then determine from that which ones is the closest to the new observation.
I know that kNN is a time comsuming process, but is this really the reason.
It seems like a too simple problem?
Aucun commentaire:
Enregistrer un commentaire