/// <summary>
/// Silhouette width:
/// s(i) = b(i) - a(i) / max(a(i), b(i))
/// d(i, c) = avg(all j in cluster c)( | x(i) - x(j) |)
/// a(i) = d(i, c) where c is i's cluster
/// b(i) = lowest d(i, c) for all c which aren't i's cluster
/// f**k it, just use R.
/// actually ballocks, can't, R can't handle big matrices
/// </summary>
public static void CalculateSilhouette(ResultClusterer.ForStat ass, IReadOnlyList <Cluster> clusters, out double silhouette, out Cluster nearestCluster)
{
// a = d(i, c) where c is i's cluster
double a = CalculateSilhouette_CalculateD(ass, ass.Assignment.Cluster.Assignments.List);
Cluster nearest = null;
double b = double.MaxValue;
foreach (Cluster cluster in clusters)
{
if (cluster != ass.Assignment.Cluster)
{
double d = CalculateSilhouette_CalculateD(ass, cluster.Assignments.List);
if (d < b)
{
b = d;
nearest = cluster;
}
}
}
double s = (b - a) / Math.Max(a, b);
silhouette = s;
nearestCluster = nearest;
}
/// <summary>
/// Used by CalculateSilhouette.
/// Calculates d(i).
/// </summary>
private static double CalculateSilhouette_CalculateD(ResultClusterer.ForStat stat, IReadOnlyList <Assignment> ass2s)
{
double[] result = new double[ass2s.Count];
for (int index = 0; index < ass2s.Count; index++)
{
result[index] = stat.DistanceMatrix.Values[stat.Assignment.Vector.Index, ass2s[index].Vector.Index];
}
if (result.Length == 0)
{
return(double.NaN);
}
return(result.Average());
}
