This package implements density-based spatial clustering of applications with noise DBSCAN in Golang. It is a data clustering algorithm proposed by Martin Ester, Hans-Peter Kriegel, Jörg Sander and Xiaowei Xu in 1996 which clusters densely packed points. This particular implementation is a fork of sohlich/go-dbscan with few optimizations and some clean up to make it a bit more idiomatic.
In order to use it, a point needs to implement DistanceTo and Name functions.
type Value float64
func (v Value) DistanceTo(other dbscan.Point) float64 {
return math.Abs(float64(other.(Value)) - v)
}
func (v Value) Name() string {
return fmt.Sprint(v)
}
After this, dbscan.Cluster function can be called with a list of points, desired minimum density and epsilon value.
clusters := dbscan.Cluster(2, 1.0,
Value(1),
Value(0.5),
Value(0),
Value(5),
Value(4.5),
Value(4),
)