search

davpinto / fastknn

Fast k-Nearest Neighbors Classifier for Large Datasets
https://davpinto.github.io/fastknn
67 stars 18 forks source link

Custom distance #4

Open mvt-oviedo opened 3 years ago

mvt-oviedo commented 3 years ago

Hi,

The package is very interesting. I was wondering if I can use a custom distance or custom matrix distance.

Thanks, and good job,

davpinto commented 3 years ago

Hi Manuel,

Thank you for your feedback.

Unfortunately not.. Since fastknn uses the RANN package to find nearest neighbors efficiently, it depends on distances implemented in the RANN package. So far, only Euclidean distance is available :(

I strongly agree that other distance metrics would be very useful. But I have no prediction to add them.. Sorry!

David Pinto Master's Degree in Computational Intelligence Federal University of Minas Gerais (UFMG) My Website https://sites.google.com/site/davidpintohomepage/home

Belo Horizonte, Minas Gerais, Brazil https://www.facebook.com/david.pinto.eng.aut https://twitter.com/davidengaut https://www.linkedin.com/in/davidpintoml

On Thu, Apr 22, 2021 at 2:19 PM Manuel Villacorta Tilve < @.***> wrote:

Hi,

The package is very interesting. I was wondering if I can use a custom distance or custom matrix distance.

Thanks, and good job,

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/davpinto/fastknn/issues/4, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACZXFDC4YRYZ6SRG3T3OTZDTKBLBTANCNFSM43M535RA .

mvt-oviedo commented 3 years ago

Many thanks,