static void Main(string[] args)
{
IKMeans _kMeans = new KMeans();
IDistance _distance = new EuclideanDistance();
Random rnd = new Random(0);
int objCount = 10;
int intrstsCount = 2;
int clusterCount = 2;
List <ClusterEntityDTO> clusters = new List <ClusterEntityDTO>();
for (int i = 0; i < objCount; i++)
{
ClusterEntityDTO dTO = new ClusterEntityDTO
{
Name = string.Format("{0:yyyyMMddHHmmss}", DateTime.Now),
Id = i
};
List <Interest> interests = new List <Interest>();
Console.WriteLine("___" + dTO.Name + "___");
for (int j = 0; j < intrstsCount; j++)
{
double rand = Math.Round(rnd.NextDouble() * 200.0);
Interest intr = new Interest
{
Name = string.Format("intr{0:yyyyMMddHHmmss}", DateTime.Now),
Weight = rand,
Weight2 = rand
};
Console.WriteLine("--- " + intr.Weight + "---");
interests.Add(intr);
}
Console.WriteLine("-------------------------------");
dTO.Interests = interests;
clusters.Add(dTO);
}
Console.WriteLine("-----------------------------------");
Console.WriteLine("-----------------------------------");
Console.WriteLine("-----------------------------------");
var result = _kMeans.Clustering(clusters, _distance, clusterCount);
for (int i = 0; i < objCount; i++)
{
Console.WriteLine("___" + result[i].Name + "___");
for (int j = 0; j < intrstsCount; j++)
{
Console.WriteLine("---" + result[i].Interests[j].Weight2 + "---");
}
if (i < objCount - 1 && result[i].CentroidId != result[i + 1].CentroidId)
{
Console.WriteLine("%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%");
}
}
Console.ReadLine();
}
public double GetDinstance(ClusterEntityDTO first, ClusterEntityDTO second)
{
double distance = 0.0;
for (int i = 0; i < first.Interests.Count; i++)
{
double diff = first.Interests[i].Weight - second.Interests[i].Weight;
distance += Math.Abs(diff);
}
return(distance);
}
public double GetDistance(List <ClusterEntityDTO> first, List <ClusterEntityDTO> second, IDistance distance)
{
ClusterEntityDTO firstDTO = new ClusterEntityDTO();
ClusterEntityDTO secondDTO = new ClusterEntityDTO();
double dist = 0.0;
for (int j = 0; j < first[0].Interests.Count; j++)
{
double weight = 0.0;
for (int i = 0; i < first.Count; i++)
{
weight += first[i].Interests[j].Weight;
}
firstDTO.Interests.Add(new Interest
{
Name = first[0].Interests[j].Name,
Weight = weight / (double)first.Count
});
}
for (int j = 0; j < second[0].Interests.Count; j++)
{
double weight = 0.0;
for (int i = 0; i < second.Count; i++)
{
weight += second[i].Interests[j].Weight;
}
secondDTO.Interests.Add(new Interest
{
Name = first[0].Interests[j].Name,
Weight = weight / second.Count
});
}
dist = distance.GetDinstance(firstDTO, secondDTO);
return(dist);
}
private List <ClusterEntityDTO> SetCentroids(ref List <ClusterEntityDTO> clusters, int clustersCount)
{
Random rnd = new Random(0);
List <ClusterEntityDTO> centroids = new List <ClusterEntityDTO>();
for (int i = 0; i < clustersCount; i++)
{
ClusterEntityDTO dTO = clusters[rnd.Next(0, clusters.Count - 1)];
if (centroids.Contains(dTO))
{
i--;
}
else
{
dTO.Id = i;
centroids.Add(dTO.CloneObjectSerializable());
}
}
return(centroids);
}
private List <ClusterEntityDTO> GetDataFromTxt(IFormFile file)
{
List <string> lines = new List <string>();
using (StreamReader reader = new StreamReader(file.OpenReadStream()))
{
string line;
while ((line = reader.ReadLine()) != null)
{
lines.Add(line);
}
}
string[] names = lines[0].Split("\t");
List <ClusterEntityDTO> result = new List <ClusterEntityDTO>();
for (int i = 1; i < lines.Count; i++)
{
ClusterEntityDTO dTO = new ClusterEntityDTO();
string[] parametrs = lines[i].Split("\t");
dTO.Interests = new List <Interest>();
dTO.Name = parametrs[0];
for (int j = 1; j < parametrs.Length; j++)
{
Interest intrst = new Interest
{
Name = names[j],
Weight = double.Parse(parametrs[j]),
Weight2 = double.Parse(parametrs[j])
};
dTO.Interests.Add(intrst);
}
result.Add(dTO);
}
return(result);
}
private List <ClusterEntityDTO> GetDataFromExcel(IFormFile file)
{
var wb = new XLWorkbook(file.OpenReadStream());
var ws = wb.Worksheets.First();
var range = ws.RangeUsed();
var rowCount = range.RowCount();
var colCount = range.ColumnCount();
List <ClusterEntityDTO> result = new List <ClusterEntityDTO>();
List <string> names = new List <string>();
for (int i = 2; i <= colCount; i++)
{
names.Add(ws.Cell(1, i).Value.ToString());
}
for (int i = 2; i <= rowCount; i++)
{
ClusterEntityDTO dTO = new ClusterEntityDTO();
dTO.Name = ws.Cell(i, 1).Value.ToString();
dTO.Interests = new List <Interest>();
for (int j = 2; j <= colCount; j++)
{
Interest interest = new Interest
{
Name = names[j - 2],
Weight = double.Parse(ws.Cell(i, j).Value.ToString()),
Weight2 = double.Parse(ws.Cell(i, j).Value.ToString())
};
dTO.Interests.Add(interest);
}
result.Add(dTO);
}
return(result);
}
private List <ClusterEntityDTO> Union(List <ClusterEntityDTO> data, int maxUnion, int step)
{
int unionCount = maxUnion;
List <ClusterEntityDTO> result = new List <ClusterEntityDTO>();
List <HierarchicalUnionObjectsDTO> distances = new List <HierarchicalUnionObjectsDTO>();
for (int i = 0; i < data.Count; i++)
{
for (int j = i + 1; j < data.Count; j++)
{
distances.Add(new HierarchicalUnionObjectsDTO
{
FirstId = i,
SecondId = j,
Distance = _distance.GetDinstance(data[i], data[j])
});
}
}
distances.Sort((x, y) => x.Distance.CompareTo(y.Distance));
bool[] used = new bool[data.Count];
int counter = 0;
for (int i = 0; i < distances.Count; i++)
{
if (unionCount == 0)
{
break;
}
if (!used[distances[i].FirstId] && !used[distances[i].SecondId])
{
ClusterEntityDTO dTO = new ClusterEntityDTO();
dTO.Children = new List <ClusterEntityDTO>
{
data[distances[i].FirstId],
data[distances[i].SecondId]
};
if (step == 1)
{
dTO.Name = GetName(counter++);
}
else
{
dTO.Name = dTO.Children[0].Name + dTO.Children[1].Name;
}
used[distances[i].FirstId] = true;
used[distances[i].SecondId] = true;
result.Add(dTO);
--unionCount;
}
}
for (int i = 0; i < data.Count; i++)
{
if (!used[i])
{
ClusterEntityDTO temp = new ClusterEntityDTO(data[i]);
temp.Children = new List <ClusterEntityDTO>(new List <ClusterEntityDTO> {
data[i]
});
temp.Name = data[i].Children == null?GetName(counter ++) : data[i].Name;
result.Add(temp);
//result.Add(data[i]);
}
}
for (int k = 0; k < result.Count; k++)
{
result[k].Interests = new List <Interest>(data[0].Interests);
if (result[k].Children != null)
{
for (int j = 0; j < result[k].Children[0].Interests.Count; j++)
{
double weight = 0.0;
for (int i = 0; i < result[k].Children.Count; i++)
{
weight += result[k].Children[i].Interests[j].Weight;
}
result[k].Interests[j].Weight = weight / result[k].Children.Count;
}
}
}
return(result);
}
