public override bool Equals(object obj)
{
if (obj == null || GetType() != obj.GetType())
{
return(false);
}
HierarchicalClusterNode other = (HierarchicalClusterNode)obj;
bool sameOrFlipped = ((left == other.left) && (right == other.right)) ||
((right == other.left) && (left == other.right));
return(sameOrFlipped && (Math.Abs(distance - other.distance) < 0.0001));
}
//TODO: do this without recusions
private static int CalcSizes(IList <HierarchicalClusterNode> nodes, IList <int> sizes, int i)
{
if (i < 0 || i >= nodes.Count)
{
return(0);
}
HierarchicalClusterNode node = nodes[i];
int leftSize = node.left >= 0 ? 1 : CalcSizes(nodes, sizes, -node.left - 1);
int rightSize = node.right >= 0 ? 1 : CalcSizes(nodes, sizes, -node.right - 1);
sizes[i] = leftSize + rightSize;
return(leftSize + rightSize);
}
/// <summary>
/// Utility format for reading clustering results from R.
/// </summary>
/// <param name="left">first column of <code>hclust$merge</code></param>
/// <param name="right">second column of <code>hclust$merge</code></param>
/// <param name="distance"><code>hclust$height</code></param>
/// <returns></returns>
public static HierarchicalClusterNode[] FromRFormat(int[] left, int[] right, double[] distance)
{
int n = distance.Length;
HierarchicalClusterNode[] nodes = new HierarchicalClusterNode[n];
for (int i = 0; i < n; i++)
{
nodes[i] = new HierarchicalClusterNode {
distance = distance[i],
left = left[i] < 0 ? -left[i] - 1 : -left[i],
right = right[i] < 0 ? -right[i] - 1 : -right[i]
};
}
return(nodes);
}
private static HierarchicalClusterNode[] FillTerminalBranch(IList <int> inds, int firstInd)
{
HierarchicalClusterNode[] result = new HierarchicalClusterNode[inds.Count - 1];
result[0] = new HierarchicalClusterNode {
left = inds[0], right = inds[1]
};
for (int i = 1; i < result.Length; i++)
{
int nodeInd = firstInd + i - 1;
result[i] = new HierarchicalClusterNode {
left = inds[i + 1], right = -1 - nodeInd
};
}
return(result);
}
/// <summary>
///
/// </summary>
/// <param name="nodes">The cluster nodes serve as input here.</param>
/// <param name="sizes"></param>
/// <param name="start"></param>
/// <param name="end"></param>
/// <param name="itemOrder"></param>
/// <param name="itemOrderInv"></param>
public static void CalcTree(HierarchicalClusterNode[] nodes, out int[] sizes, out int[] start, out int[] end,
out int[] itemOrder, out int[] itemOrderInv)
{
if (nodes == null){
sizes = null;
start = null;
end = null;
itemOrder = null;
itemOrderInv = null;
return;
}
sizes = new int[nodes.Length];
start = new int[nodes.Length];
end = new int[nodes.Length];
itemOrder = new int[nodes.Length + 1];
int count = 0;
CalcItemOrder(nodes, itemOrder, nodes.Length - 1, ref count);
CalcSizes(nodes, sizes, nodes.Length - 1);
CalcStartEnd(nodes, sizes, start, end, nodes.Length - 1, 0, nodes.Length + 1);
itemOrderInv = InvertOrder(itemOrder);
}
//TODO: do this without recusions
private static void CalcStartEnd(IList <HierarchicalClusterNode> nodes, IList <int> sizes, IList <int> start,
IList <int> end, int i, int s, int e)
{
if (i == -1)
{
return;
}
start[i] = s;
end[i] = e;
HierarchicalClusterNode node = nodes[i];
int size = sizes[i];
int leftSize = node.left >= 0 ? 1 : sizes[-1 - node.left];
int rightSize = size - leftSize;
if (leftSize > 1)
{
CalcStartEnd(nodes, sizes, start, end, -1 - node.left, s, s + leftSize);
}
if (rightSize > 1)
{
CalcStartEnd(nodes, sizes, start, end, -1 - node.right, s + leftSize, e);
}
}
public HierarchicalClusterNode[] TreeClusterKmeans(MatrixIndexer data, MatrixAccess access, IDistance distance,
HierarchicalClusterLinkage linkage, bool preserveOrder, bool periodic, int nthreads, int nmeans, int restarts,
int maxIter, Action <int> progress)
{
int nelements = (access == MatrixAccess.Rows) ? data.RowCount : data.ColumnCount;
if (nelements <= nmeans)
{
return(TreeCluster(data, access, distance, linkage, preserveOrder, periodic, nthreads, progress));
}
float[,] c;
int[] inds;
if (access == MatrixAccess.Rows)
{
KmeansClustering.GenerateClusters(data, nmeans, maxIter, restarts, progress, out c, out inds);
}
else
{
KmeansClustering.GenerateClusters(data.Transpose(), nmeans, maxIter, restarts, progress, out c, out inds);
}
float[,] distMatrix = DistanceMatrix(new FloatMatrixIndexer(c), distance, MatrixAccess.Rows);
HierarchicalClusterNode[] nodes = TreeCluster(distMatrix, linkage, preserveOrder, periodic, nthreads, progress);
Dictionary <int, int[]> clusters;
Dictionary <int, int> singletons;
RearrangeClusters(inds, c.GetLength(0), out clusters, out singletons);
HierarchicalClusterNode[] newNodes = new HierarchicalClusterNode[nelements - 1];
int fill = nelements - nmeans;
Array.Copy(nodes, 0, newNodes, fill, nodes.Length);
int pos = 0;
for (int i = fill; i < newNodes.Length; i++)
{
HierarchicalClusterNode node = newNodes[i];
if (node.left < 0)
{
node.left -= fill;
}
else if (singletons.ContainsKey(node.left))
{
node.left = singletons[node.left];
}
else
{
if (clusters.ContainsKey(node.left))
{
HierarchicalClusterNode[] branch = FillTerminalBranch(clusters[node.left], pos);
Array.Copy(branch, 0, newNodes, pos, branch.Length);
pos += branch.Length;
node.left = -pos;
}
}
if (node.right < 0)
{
node.right -= fill;
}
else if (singletons.ContainsKey(node.right))
{
node.right = singletons[node.right];
}
else
{
if (clusters.ContainsKey(node.right))
{
HierarchicalClusterNode[] branch = FillTerminalBranch(clusters[node.right], pos);
Array.Copy(branch, 0, newNodes, pos, branch.Length);
pos += branch.Length;
node.right = -pos;
}
}
}
return(newNodes);
}
private static HierarchicalClusterNode[] FillTerminalBranch(IList<int> inds, int firstInd)
{
HierarchicalClusterNode[] result = new HierarchicalClusterNode[inds.Count - 1];
result[0] = new HierarchicalClusterNode{left = inds[0], right = inds[1]};
for (int i = 1; i < result.Length; i++){
int nodeInd = firstInd + i - 1;
result[i] = new HierarchicalClusterNode{left = inds[i + 1], right = -1 - nodeInd};
}
return result;
}
public HierarchicalClusterNode[] TreeClusterKmeans(MatrixIndexer data, MatrixAccess access, IDistance distance,
HierarchicalClusterLinkage linkage, bool preserveOrder, bool periodic, int nthreads, int nmeans, int restarts,
int maxIter, Action<int> progress)
{
int nelements = (access == MatrixAccess.Rows) ? data.RowCount : data.ColumnCount;
if (nelements <= nmeans){
return TreeCluster(data, access, distance, linkage, preserveOrder, periodic, nthreads, progress);
}
float[,] c;
int[] inds;
if (access == MatrixAccess.Rows){
KmeansClustering.GenerateClusters(data, nmeans, maxIter, restarts, progress, out c, out inds);
} else{
KmeansClustering.GenerateClusters(data.Transpose(), nmeans, maxIter, restarts, progress, out c, out inds);
}
float[,] distMatrix = DistanceMatrix(new FloatMatrixIndexer(c), distance, MatrixAccess.Rows);
HierarchicalClusterNode[] nodes = TreeCluster(distMatrix, linkage, preserveOrder, periodic, nthreads, progress);
Dictionary<int, int[]> clusters;
Dictionary<int, int> singletons;
RearrangeClusters(inds, c.GetLength(0), out clusters, out singletons);
HierarchicalClusterNode[] newNodes = new HierarchicalClusterNode[nelements - 1];
int fill = nelements - nmeans;
Array.Copy(nodes, 0, newNodes, fill, nodes.Length);
int pos = 0;
for (int i = fill; i < newNodes.Length; i++){
HierarchicalClusterNode node = newNodes[i];
if (node.left < 0){
node.left -= fill;
} else if (singletons.ContainsKey(node.left)){
node.left = singletons[node.left];
} else{
if (clusters.ContainsKey(node.left)){
HierarchicalClusterNode[] branch = FillTerminalBranch(clusters[node.left], pos);
Array.Copy(branch, 0, newNodes, pos, branch.Length);
pos += branch.Length;
node.left = -pos;
}
}
if (node.right < 0){
node.right -= fill;
} else if (singletons.ContainsKey(node.right)){
node.right = singletons[node.right];
} else{
if (clusters.ContainsKey(node.right)){
HierarchicalClusterNode[] branch = FillTerminalBranch(clusters[node.right], pos);
Array.Copy(branch, 0, newNodes, pos, branch.Length);
pos += branch.Length;
node.right = -pos;
}
}
}
return newNodes;
}
