Plot CreateIntraClusterPlot(ModelAnalyzer analyzer)
{
Plot intraClusterPlot = new Plot();
int pointsCount = analyzer.AnalyzedSongs.Count;
double[] xValues = new double[pointsCount];
double[] yValues = new double[pointsCount];
int currentClusterIndex = analyzer.AnalyzedSongs[0].ClusterIndex;
for (int i = 0; i < pointsCount; i++)
{
AnalyzedSong point = analyzer.AnalyzedSongs[i];
xValues[i] = i;
yValues[i] = point.CentroidDistance;
if (currentClusterIndex != point.ClusterIndex)
{
double value = (i + (i - 1)) / 2.0;
string label = $"cluster {currentClusterIndex}";
intraClusterPlot.PlotVLine(value, VerticalDividerColor, label: label);
}
currentClusterIndex = point.ClusterIndex;
}
intraClusterPlot.PlotScatter(xValues, yValues, IntraClusterDistanceColor);
intraClusterPlot.PlotHLine(analyzer.IntraClusterMeanDistance, IntraClustersMeanDistanceColor);
return(intraClusterPlot);
}
void AnalyzeIntraClusterDistance()
{
List <Song> centroids = Model.Centroids;
IntraClusterDistances = new List <double>(centroids.Capacity);
IntraClusterMeanDistance = 0.0;
List <int> clusterCounts = new List <int>(centroids.Capacity);
int songsCount = 0;
for (int i = 0; i < centroids.Count; i++)
{
IntraClusterDistances.Add(0.0);
clusterCounts.Add(0);
}
AnalyzedSongs = Model.ClusterizedSongs.Select(song =>
{
double distance = Model.DistanceFunc.GetDistance(song.Song, centroids[song.ClusterIndex]);
int index = song.ClusterIndex;
AnalyzedSong analyzedSong = new AnalyzedSong
{
AssociatedSong = song.Song,
ClusterIndex = index,
CentroidDistance = distance,
};
IntraClusterDistances[index] += distance;
IntraClusterMeanDistance += distance;
clusterCounts[index] += 1;
songsCount += 1;
return(analyzedSong);
}).ToList();
for (int i = 0; i < IntraClusterDistances.Count; i++)
{
IntraClusterDistances[i] /= clusterCounts[i];
}
IntraClusterMeanDistance /= songsCount;
}
