tjkemper / knn

k-nearest neighbors (knn) implemented with a k-d tree

knn with k-d trees

Terms

k-nearest neighbors (knn)
k-dimensional tree (k-d tree)

Description

The k-d tree data structure allows for efficient knn algorithms. A simple implementation can be found in kdtree.py.

Visualizing k-d trees and knn

2D

3D

KDTree usage

Initialize

num_dims = 2
points = [
    [0, 0],
    [4, 6],
    [5, 1],
    [6, -6],
    [-5, -5],
    [-4, -1],
    [-3, 5],
]
tree = KDTree(points, num_dims)

Visualize

tree.visualize(visual_type=VisualType.textual)
tree.visualize(visual_type=VisualType.graphical)

knn

point = [1, 4]
k = 3
result = tree.knn(point, k)
tree.visualize_knn(point, result)