public object Clone()
{
Cluster pRet = new Cluster();
pRet.m_index = this.m_index;
pRet.m_name = this.m_name;
pRet.m_center = (double[])this.Center.Clone();
if (this.m_elements != null)
{
pRet.m_elements = new List<IndivData>();
foreach (var p in this.m_elements)
{
pRet.m_elements.Add((IndivData)p.Clone());
}
}
return pRet;
}
private bool initializeCiusters(int nbClusters)
{
bool oRet = false;
var inds = this.Indivs;
if ((inds == null) || (nbClusters < 2))
{
return oRet;
}
if (inds.Count < nbClusters)
{
return oRet;
}
double[] dd = inds.First().DoubleData;
int nv = dd.Length;
if (nv < 1)
{
return oRet;
}
double[,] dLimits = new double[3, nv];
for (int i = 0; i < nv; ++i)
{
double x = dd[i];
dLimits[0, i] = x;
dLimits[1, i] = x;
dLimits[2, i] = 0.0;
}// i
int nc = 0;
foreach (var ind in inds)
{
double[] ddx = ind.DoubleData;
if (ddx.Length < nv)
{
continue;
}
++nc;
for (int i = 0; i < nv; ++i)
{
double x = ddx[i];
if (x < dLimits[0, i])
{
dLimits[0, i] = x;
}
else if (x > dLimits[1, i])
{
dLimits[1, i] = x;
}
}// i
}// ind
if (nc < 1)
{
return oRet;
}
for (int i = 0; i < nv; ++i)
{
double x1 = dLimits[0, i];
double x2 = dLimits[1, i];
double step = (x2 - x1) / (nbClusters + 2);
if (step <= 0.0)
{
return oRet;
}
dLimits[2, i] = step;
}// i
Cluster[] cc = new Cluster[nbClusters];
for (int i = 0; i < nbClusters; ++i)
{
double[] xc = new double[nv];
for (int j = 0; j < nv; ++j)
{
xc[j] = dLimits[0, j] + (i + 1) * dLimits[2, j];
}// j
Cluster c = new Cluster();
c.Index = i;
c.Name = String.Format("KM{0}", i + 1);
c.Center = xc;
cc[i] = c;
}// i
this.m_clusters = cc.ToList();
return true;
}
public static CategClusterSet ProcessAsync(OrdModelView model, int nClusters, CancellationToken cancellationToken)
{
if ((model == null) || (nClusters < 2))
{
return null;
}
if (cancellationToken.IsCancellationRequested)
{
return null;
}
var rd = new Random();
CategClusterSet oRet = new CategClusterSet(model);
Cluster[] cc = new Cluster[nClusters];
IndivData[] inds = oRet.Indivs.ToArray();
int nMax = inds.Length;
if (nMax < nClusters)
{
return null;
}
HashSet<int> oCur = new HashSet<int>();
for (int i = 0; i < nClusters; ++i)
{
int nx = rd.Next(0, nMax);
while (nx >= nMax)
{
nx = rd.Next(nMax);
}
IndivData indiv = inds[nx];
int index = indiv.IndivIndex;
while (oCur.Contains(index))
{
nx = rd.Next(0, nMax);
while (nx >= nMax)
{
nx = rd.Next(0, nMax);
}
indiv = inds[nx];
index = indiv.IndivIndex;
}
oCur.Add(index);
var xc = new Cluster(indiv, ClassificationType.Utility);
xc.Name = String.Format("CU{0}", i + 1);
xc.Index = i;
cc[i] = xc;
}// i
oRet.Clusters = cc.ToList();
double xx = oRet.compute_global_crit(cancellationToken);
oRet.GlobalCrit = xx;
if (cancellationToken.IsCancellationRequested)
{
return null;
}
foreach (var ind in inds)
{
if (cancellationToken.IsCancellationRequested)
{
return null;
}
if (!oCur.Contains(ind.IndivIndex))
{
if (!oRet.processOne(ind, cancellationToken))
{
return null;
}
}
}// inds
return oRet;
}
private bool contains(Cluster oCluster)
{
Cluster cluster = null;
foreach (var ind1 in oCluster.Elements)
{
if (cluster != null)
{
break;
}
int nIndex = ind1.IndivIndex;
foreach (var c in this.Clusters)
{
var q = from x in c.Elements where x.IndivIndex == nIndex select x;
if (q.Count() > 0)
{
cluster = c;
break;
}
}// c
}// ind1
if (cluster == null)
{
return false;
}
if (cluster.Count != oCluster.Count)
{
return false;
}
foreach (var ind1 in oCluster.Elements)
{
var q = from x in cluster.Elements where x.IndivIndex == ind1.IndivIndex select x;
if (q.Count() < 1)
{
return false;
}
}// ind1
return true;
}
public static Task<CategClusterSet> KMeansAsync(OrdModelView model, int nClusters, int nbIterations, CancellationToken cancellationToken,
IProgress<Tuple<int, CategClusterSet>> progress)
{
return Task.Run<CategClusterSet>(() =>
{
CategClusterSet oBest = new CategClusterSet(model);
oBest.initializeCiusters(nClusters);
for (int i = 0; i < nbIterations; ++i)
{
CategClusterSet pp = new CategClusterSet();
pp.m_model = oBest.m_model;
pp.m_nvars = oBest.m_nvars;
pp.m_indivs = oBest.m_indivs;
List<Cluster> oList = new List<Cluster>();
foreach (var c in oBest.Clusters)
{
Cluster cc = new Cluster();
cc.Index = c.Index;
cc.Name = c.Name;
cc.Center = (double[])c.Center.Clone();
oList.Add(cc);
}// c
pp.Clusters = oList;
int f = (int)(((double)i / (double)nbIterations) * 100.0 + 0.5);
if (cancellationToken.IsCancellationRequested)
{
break;
}
pp.kmeansstep(cancellationToken);
if (cancellationToken.IsCancellationRequested)
{
break;
}
if (oBest.Equals(pp))
{
break;
}
oBest = pp;
if (progress != null)
{
progress.Report(new Tuple<int, CategClusterSet>(f, pp));
}
}// i
return oBest;
}, cancellationToken);
}
