README
¶
K-Means
kmeans is BSD licensed fast k-means implementation(with better initialization, known as k-means++) written in Go. It supports various distance functions out of the box for convenience and experimentation. It has large coverage for tests. The algorithm is tested on Iris dataset and distance functions have full test coverage.
Documentation
License
Distance Functions
It supports various distance functions:
- LP norms (manhattan, euclidean distances including)
- SquaredEuclideanDistance
- Minkowski Distance
- Weighted Minkowski Distance
- Chebyshev Distance
- Hamming Distance
- Bray Curtis Distance
- CanberraDistance
Documentation
¶
Index ¶
- func BrayCurtisDistance(firstVector, secondVector []float64) (float64, error)
- func CanberraDistance(firstVector, secondVector []float64) (float64, error)
- func ChebyshevDistance(firstVector, secondVector []float64) (float64, error)
- func EuclideanDistance(firstVector, secondVector []float64) (float64, error)
- func HammingDistance(firstVector, secondVector []float64) (float64, error)
- func Kmeans(rawData [][]float64, k int, distanceFunction DistanceFunction, threshold int) ([]int, error)
- func LPNorm(vector []float64, p float64) (float64, error)
- func ManhattanDistance(firstVector, secondVector []float64) (float64, error)
- func MinkowskiDistance(firstVector, secondVector []float64, p float64) (float64, error)
- func SquaredEuclideanDistance(firstVector, secondVector []float64) (float64, error)
- func WeightedMinkowskiDistance(firstVector, secondVector, weightVector []float64, p float64) (float64, error)
- type ClusteredObservation
- type DistanceFunction
- type Observation
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func BrayCurtisDistance ¶
func CanberraDistance ¶
func ChebyshevDistance ¶
infinity norm distance (l_inf distance)
func EuclideanDistance ¶
2-norm distance (l_2 distance)
func HammingDistance ¶
func Kmeans ¶
func Kmeans(rawData [][]float64, k int, distanceFunction DistanceFunction, threshold int) ([]int, error)
K-Means Algorithm with smart seeds as known as K-Means ++
func ManhattanDistance ¶
1-norm distance (l_1 distance)
func MinkowskiDistance ¶
p-norm distance (l_p distance)
func SquaredEuclideanDistance ¶
Higher weight for the points that are far apart Not a real metric as it does not obey triangle inequality
Types ¶
type ClusteredObservation ¶
type ClusteredObservation struct {
ClusterNumber int
Observation
}
Abstracts the Observation with a cluster number Update and computeation becomes more efficient
type DistanceFunction ¶
Distance Function: To compute the distanfe between observations
type Observation ¶
type Observation []float64
Observation: Data Abstraction for an N-dimensional observation
func (Observation) Add ¶
func (observation Observation) Add(otherObservation Observation)
Summation of two vectors
func (Observation) InnerProduct ¶
func (observation Observation) InnerProduct(otherObservation Observation)
Dot Product of Two vectors
func (Observation) Mul ¶
func (observation Observation) Mul(scalar float64)
Multiplication of a vector with a scalar
func (Observation) OuterProduct ¶
func (observation Observation) OuterProduct(otherObservation Observation) [][]float64
Outer Product of two arrays TODO: Need to be tested
Source Files