Classification 2 classes and 60 Querypoints

Rookfighter / knn-cpp

A header-only C++ library for k nearest neighbor search with Eigen3.

MIT License

51 stars 5 forks source link

Classification 2 classes and 60 Querypoints #3

Closed Youssef-Zarca closed 2 years ago

Youssef-Zarca commented 3 years ago

I have 2 classes and I want to see if the 60 points are one of these 2 classes. The problem is that I can’t get much data when I implement this algorithm with more than 1 nearest neighbor, cause I get a lot of -1 output otherwise.

please how can I use this algorithm with 2 classes and 60 points to be classified with k=7 without getting -1 in the output? Please can I iterate in the algorithm and adding points to the classes to be looked for as neighbors or the algorithm do that automatically?

Rookfighter commented 3 years ago

Hi and thanks for using knn-cpp. I am happy to help you as best as I can. First off, can you give me a bit more details on your data? Like:

What dimensionality do your points have?
What data do you use to build the KDTree?
What data do you use to query the KDTree?

Maybe you can also post some example code on how you use the KDTree, so I can get more insights into your problem.

Youssef-Zarca commented 3 years ago

Thank you so much for your reply as well. Actually I have solved the problem. I have used the library wrong. Now my Datapoints are the EEG Chanels Features and the query points are the reference features, which I need to look for their next k Neighbors.

The KDTree is the one in the readme file with Euclidean distance. From the distances Matrix I do the classification.

thank you deeply