static void Main(string[] args)
{
MyCustomDatasetItem[] featureData = { };
List<MyCustomDatasetItem> testPoints = new List<MyCustomDatasetItem>();
for (int i = 0; i < 1000; i++)
{
//points around (1,1) with most 1 distance
testPoints.Add(new MyCustomDatasetItem(1, 1 + ((float)i / 1000)));
testPoints.Add(new MyCustomDatasetItem(1, 1 - ((float)i / 1000)));
testPoints.Add(new MyCustomDatasetItem(1 - ((float)i / 1000), 1));
testPoints.Add(new MyCustomDatasetItem(1 + ((float)i / 1000), 1));
//points around (5,5) with most 1 distance
testPoints.Add(new MyCustomDatasetItem(5, 5 + ((float)i / 1000)));
testPoints.Add(new MyCustomDatasetItem(5, 5 - ((float)i / 1000)));
testPoints.Add(new MyCustomDatasetItem(5 - ((float)i / 1000), 5));
testPoints.Add(new MyCustomDatasetItem(5 + ((float)i / 1000), 5));
}
featureData = testPoints.ToArray();
HashSet<MyCustomDatasetItem[]> clusters;
var dbs = new DbscanAlgorithm<MyCustomDatasetItem>((x, y) => Math.Sqrt(((x.X - y.X) * (x.X - y.X)) + ((x.Y - y.Y) * (x.Y - y.Y))));
dbs.ComputeClusterDbscan(allPoints: featureData, epsilon: .01, minPts: 10, clusters: out clusters);
}
/// <summary>
/// Most basic usage of Dbscan implementation, with no eventing and async mechanism
/// </summary>
/// <param name="features">Features provided</param>
private static void RunOfflineDbscan(List <MyFeature> features)
{
var simpleDbscan = new DbscanAlgorithm <MyFeature>(EuclidienDistance);
var result = simpleDbscan.ComputeClusterDbscan(allPoints: features.ToArray(),
epsilon: .01, minimumPoints: 10);
Console.WriteLine($"Noise: {result.Noise.Count}");
Console.WriteLine($"# of Clusters: {result.Clusters.Count}");
}
static void Main(string[] args)
{
MyCustomFeature[] featureData = { };
var testPoints = new List <MyCustomFeature>();
//points around (1,1) with most 1 distance
testPoints.Add(new MyCustomFeature(1, 1));
for (int i = 1; i <= 1000; i++)
{
float v = ((float)i / 1000);
testPoints.Add(new MyCustomFeature(1, 1 + v));
testPoints.Add(new MyCustomFeature(1, 1 - v));
testPoints.Add(new MyCustomFeature(1 - v, 1));
testPoints.Add(new MyCustomFeature(1 + v, 1));
}
//points around (5,5) with most 1 distance
testPoints.Add(new MyCustomFeature(5, 5));
for (int i = 1; i <= 1000; i++)
{
float v = ((float)i / 1000);
testPoints.Add(new MyCustomFeature(5, 5 + v));
testPoints.Add(new MyCustomFeature(5, 5 - v));
testPoints.Add(new MyCustomFeature(5 - v, 5));
testPoints.Add(new MyCustomFeature(5 + v, 5));
}
featureData = testPoints.ToArray();
//INFO: applied euclidean distance as metric calculation function
var dbscan = new DbscanAlgorithm <MyCustomFeature>(
(feature1, feature2) =>
Math.Sqrt(
((feature1.X - feature2.X) * (feature1.X - feature2.X)) +
((feature1.Y - feature2.Y) * (feature1.Y - feature2.Y))
)
);
var result = dbscan.ComputeClusterDbscan(allPoints: featureData, epsilon: .01, minimumPoints: 10);
}
/// <summary>
/// This is the method that actually does the work.
/// </summary>
/// <param name="DA">The DA object is used to retrieve from inputs and store in outputs.</param>
protected override void SolveInstance(IGH_DataAccess DA)
{
bool IsPointData = false;
GH_Structure <IGH_Goo> data = new GH_Structure <IGH_Goo>();
List <double> eps = new List <double>();
List <int> minP = new List <int>();
if (!DA.GetDataTree(0, out data))
{
return;
}
if (!DA.GetDataList(1, eps))
{
return;
}
if (!DA.GetDataList(2, minP))
{
return;
}
data.Simplify(GH_SimplificationMode.CollapseAllOverlaps);
List <DataSetItemPoint[]> points = new List <DataSetItemPoint[]>();
for (int i = 0; i < data.Branches.Count; i++)
{
DataSetItemPoint[] pp = new DataSetItemPoint[data.Branches[i].Count];
for (int j = 0; j < data.Branches[i].Count; j++)
{
if (data.Branches[i][j] is GH_Point)
{
IsPointData = true;
GH_Point target = new GH_Point();
if (GH_Convert.ToGHPoint(data.Branches[i][j], GH_Conversion.Both, ref target))
{
pp[j] = new DataSetItemPoint(target.Value.X, target.Value.Y, target.Value.Z, 0.0);
}
}
else
{
break;
}
}
if (IsPointData)
{
points.Add(pp);
}
else
{
break;
}
}
// double data
if (!IsPointData)
{
DataSetItemPoint[] pp = new DataSetItemPoint[data.Branches.Count];
for (int i = 0; i < data.Branches.Count; i++)
{
DataSetItemPoint p = new DataSetItemPoint();
for (int j = 0; j < data.Branches[i].Count; j++)
{
if (data.Branches[i][j] is GH_Number)
{
if (GH_Convert.ToDouble(data.Branches[i][j], out double value, GH_Conversion.Both))
{
switch (j)
{
case 0:
p.X = value;
break;
case 1:
p.Y = value;
break;
case 2:
p.Z = value;
break;
case 3:
p.W = value;
break;
}
}
}
}
pp[i] = p;
}
points.Add(pp);
}
if (IsPointData)
{
DataTree <IGH_Goo> output = new DataTree <IGH_Goo>();
for (int i = 0; i < points.Count; i++)
{
DbscanAlgorithm <DataSetItemPoint> dbs = new DbscanAlgorithm <DataSetItemPoint>((x, y) => Math.Sqrt(((x.X - y.X) * (x.X - y.X)) + ((x.Y - y.Y) * (x.Y - y.Y)) + ((x.Z - y.Z) * (x.Z - y.Z)) + ((x.W - y.W) * (x.W - y.W))));
dbs.ComputeClusterDbscan(points[i].ToArray(), eps[i], minP[i], out HashSet <DataSetItemPoint[]> clusters3d);
for (int j = 0; j < clusters3d.Count; j++)
{
ConcurrentQueue <GH_Point> _points = new ConcurrentQueue <GH_Point>();
Parallel.ForEach(clusters3d.ElementAt(j), p =>
{
_points.Enqueue(new GH_Point(new Point3d(p.X, p.Y, p.Z)));
});
output.AddRange(_points.ToList(), new GH_Path(i, j));
}
}
DA.SetDataTree(0, output);
}
else
{
DataTree <GH_Number> output = new DataTree <GH_Number>();
for (int i = 0; i < points.Count; i++)
{
DbscanAlgorithm <DataSetItemPoint> dbs = new DbscanAlgorithm <DataSetItemPoint>((x, y) => Math.Sqrt(((x.X - y.X) * (x.X - y.X)) + ((x.Y - y.Y) * (x.Y - y.Y)) + ((x.Z - y.Z) * (x.Z - y.Z)) + ((x.W - y.W) * (x.W - y.W))));
dbs.ComputeClusterDbscan(points[i].ToArray(), eps[i], minP[i], out HashSet <DataSetItemPoint[]> clusters3d);
for (int j = 0; j < clusters3d.Count; j++)
{
ConcurrentQueue <List <double> > _points = new ConcurrentQueue <List <double> >();
for (int k = 0; k < clusters3d.ElementAt(j).Length; k++)
{
List <GH_Number> ii = new List <GH_Number>();
GH_Number target1 = new GH_Number();
GH_Number target2 = new GH_Number();
GH_Number target3 = new GH_Number();
GH_Number target4 = new GH_Number();
if (GH_Convert.ToGHNumber(clusters3d.ElementAt(j).ElementAt(k).X, GH_Conversion.Both, ref target1))
{
ii.Add(target1);
}
if (GH_Convert.ToGHNumber(clusters3d.ElementAt(j).ElementAt(k).Y, GH_Conversion.Both, ref target2))
{
ii.Add(target2);
}
if (GH_Convert.ToGHNumber(clusters3d.ElementAt(j).ElementAt(k).Z, GH_Conversion.Both, ref target3))
{
ii.Add(target3);
}
if (GH_Convert.ToGHNumber(clusters3d.ElementAt(j).ElementAt(k).W, GH_Conversion.Both, ref target4))
{
ii.Add(target4);
}
output.AddRange(ii, new GH_Path(i, j, k));
}
}
}
DA.SetDataTree(0, output);
}
}
static void Main(string[] args)
{
MyCustomFeature[] featureData = { };
var testPoints = new List <MyCustomFeature>()
{
};
//points around (1,1) with most 1 distance
testPoints.Add(new MyCustomFeature(1, 1));
for (int i = 1; i <= 1000; i++)
{
var v = (float)i / 1000;
testPoints.Add(new MyCustomFeature(1, 1 + v));
testPoints.Add(new MyCustomFeature(1, 1 - v));
testPoints.Add(new MyCustomFeature(1 - v, 1));
testPoints.Add(new MyCustomFeature(1 + v, 1));
}
//points around (5,5) with most 1 distance
testPoints.Add(new MyCustomFeature(5, 5));
for (int i = 1; i <= 1000; i++)
{
var v = (float)i / 1000;
testPoints.Add(new MyCustomFeature(5, 5 + v));
testPoints.Add(new MyCustomFeature(5, 5 - v));
testPoints.Add(new MyCustomFeature(5 - v, 5));
testPoints.Add(new MyCustomFeature(5 + v, 5));
}
//noise point
testPoints.Add(new MyCustomFeature(10, 10));
featureData = testPoints.ToArray();
//INFO: applied euclidean distance as metric calculation function
var dbscan = new DbscanAlgorithm <MyCustomFeature>(
(feature1, feature2) =>
Math.Sqrt(
((feature1.X - feature2.X) * (feature1.X - feature2.X)) +
((feature1.Y - feature2.Y) * (feature1.Y - feature2.Y))
)
);
var result = dbscan.ComputeClusterDbscan(allPoints: featureData, epsilon: .01, minimumPoints: 10);
Console.WriteLine($"Noise: {result.Noise.Length}");
Console.WriteLine($"# of Clusters: {result.Clusters.Count}");
//INFO: applied euclidean distance as metric calculation function
//INFO: second argument of constructor takes an instance implemented with IDbscanEventPublisher interface
var dbscanWithEventing = new DbscanAlgorithm <MyCustomFeature>(
(feature1, feature2) =>
Math.Sqrt(
((feature1.X - feature2.X) * (feature1.X - feature2.X)) +
((feature1.Y - feature2.Y) * (feature1.Y - feature2.Y))
),
new DbscanLogger()
);
var resultWithEventing = dbscanWithEventing.ComputeClusterDbscan(allPoints: featureData, epsilon: .01, minimumPoints: 10);
Console.WriteLine($"Noise: {resultWithEventing.Noise.Length}");
Console.WriteLine($"# of Clusters: {resultWithEventing.Clusters.Count}");
}