knn is a k nearest neighbor classifier library written in Clojure. It supports a variety of distance functions out of the box, listed below. It has full test coverage.
(use ‘[knn.core :refer :all])
(use ‘[knn.distance :refer :all])
(def neighbors 3)
; After loading training data and test data into vectors(the observation vectors need to be same size)
; Predictions are the vector that has the class predictions for each observation
(def predictions (predict training-data test-data manhattan-distance neighbors)After cloning the repo, in order to generate the documentation:
; For single page documentation(uberdoc.html)
lein marg
; For multi page documentation(one page per module)
lein marg --multiCopyright © 2014 Bugra Akyildiz
Distributed under the Eclipse Public License either version 1.0 or (at your option) any later version.
- I am planning to support PigPen in future.
- KD -Trees integration in order to improve the speed even further.
Adapted from Distances.jl
| Distance name | Syntax | math definition |
|---|---|---|
| Euclidean (L_2) | (euclidean-distance x y) |
sqrt(sum((x - y) .^ 2)) |
| Squared Euclidean | (squared-euclidean-distance x y) |
sum((x - y).^2) |
| Manhattan (L_1) | (manhattan-distance x y) |
sum(abs(x - y)) |
| Chebyshev | (chebyshev-distance x y) |
max(abs(x - y)) |
| Minkowski (L_P) | (minkowski-distance x y p) |
sum(abs(x - y).^p) ^ (1/p) |
| Hamming | (hamming-distance x y) |
sum(x .!= y) |
| Cosine | (cosine-distance x y) |
1 - dot(x, y) / (norm(x) * norm(y)) |
| Correlation | (correlation-distance x y) |
cosine_dist(x - mean(x), y - mean(y)) |
| Chi Squared | (chi-squared-distance x y) |
sum((x - y).^2 / (x + y)) |
| KLDivergence | (kl-divergence x y) |
sum(p .* log(p ./ q)) |
| JSDivergence | (js-divergence x y) |
KL(x, m) / 2 + KL(y, m) / 2 with m = (x + y) / 2 |
| SpanNormDist | (span-norm-distance x y) |
max(x - y) - min(x - y ) |
| BhattacharyyaDist | (bhattacharyya-distance x y) |
-log(sum(sqrt(x .* y) / sqrt(sum(x) * sum(y))) |
| HellingerDist | (hellinger-distance x y) |
sqrt(1 - sum(sqrt(x .* y) / sqrt(sum(x) * sum(y)))) |
| WeightedMinkowski | (weighted-minkowski-distance x y w p) |
sum(abs(x - y).^p .* w) ^ (1/p) |