private void button4_Click_1(object sender, EventArgs e)
{
List <string> ids = new List <string>();
List <Case> standrize_array = new List <Case>();
//get ids od cases to delete
for (int i = 0; i < dataGridView_old.SelectedRows.Count; i++)
{
ids.Add(dataGridView_old.SelectedRows[i].Cells["id"].Value.ToString());
}
// delete cases from tables
Db.del_case_id(choice.TableName, ids);
Db.del_case_id(choice.Tablename_Standardization, ids);
// re_standrize all tables
ArrayList array = Db.get_cases_as_arraylist_condition(choice.Tablename_Standardization, choice.CaseName);
standrize_array = choice.StandardizeData(array);
Db.del_data_table(choice.Tablename_Standardization);
standrize_array.Add(choice.Case_Max_Mean);
standrize_array.Add(choice.Case_Min_SD);
int num = Convert.ToInt32(choice.Clusters_Num);
for (int i = 0; i < num; i++) // recalc centroids
{
List <Case> cases_cluster = standrize_array.FindAll(
delegate(Case ca)
{
return(ca.GetFeature("cluster").GetFeatureValue().ToString() == i.ToString());
}
);
Case centroid = Kmeans.ReCalcCentroid(cases_cluster, i);
standrize_array.Add(centroid);
}
// add all after restandrize
Db.insert_data_to_standrize_caseTable_noID(choice.Tablename_Standardization, choice.get_rows_as_string(standrize_array));
dataGridView_old.DataSource = Db.get_table_contetnt(choice.TableName);
}
private void button4_Click_2(object sender, EventArgs e)
{
if (radioButton1.Checked)
{
dG_data.DataSource = Db.get_all_orderd(cmbox_case.SelectedValue.ToString(), "cluster");
dg_clusters.DataSource = Db.get_count_cluster(cmbox_case.SelectedValue.ToString());
return;
}
lstBox_Notes.Items.Add("Normalizing Data using MIn-Max Scaling");
List <Case> standardize_current_cases = mybestchoice.StandardizeData(mybestchoice.ArrayCases);
mykp = new KmeansPlus();
KmeansPlus.PointClusters myclusters;
List <KmeansPlus.PointClusters> Allmyclusters = new List <KmeansPlus.PointClusters>();
// myclusters = mykp.GetKMeansPP(standardize_current_cases,int.Parse( Clusters_num));
double min_fk = Double.MaxValue;
int min_i = 0;
for (int j = 0; j < 10; j++)
{
myclusters = new KmeansPlus.PointClusters();
myclusters = mykp.GetKMeansPP_K(standardize_current_cases);
Allmyclusters.Add(myclusters);
if (myclusters.Fk < min_fk)
{
min_i = j;
}
}
myclusters = Allmyclusters[min_i];
lstBox_Notes.Items.Add("Best Clusters Number = " + myclusters.PC.Count.ToString());
// intialize db grids
dG_data.Rows.Clear();
dG_data.Columns.Clear();
for (int j = 0; j < mybestchoice.Cases[0].GetFeatures().Count; j++)
{
Feature feature = (Feature)mybestchoice.Cases[0].GetFeatures()[j];
dG_data.Columns.Add(feature.GetFeatureName().ToString(), feature.GetFeatureName().ToString());
}
dG_data.Rows.Add(mybestchoice.Cases.Count);
int countcluster = myclusters.PC.Count;
dg_clusters.Rows.Clear();
dg_clusters.Columns.Clear();
dg_clusters.Columns.Add("cluster", "cluster");
dg_clusters.Columns.Add("count", "count");
dg_clusters.Rows.Add(countcluster + 1);
dg_clusters.Rows[0].Cells[0].Value = "عدد العناقيد الكلي";
dg_clusters.Rows[0].Cells[1].Value = countcluster;
int t = 1;
int i = 0;
foreach (List <Case> cluster in myclusters.PC.Values)
{
dg_clusters.Rows[t].Cells[0].Value = t.ToString();
dg_clusters.Rows[t].Cells[1].Value = cluster.Count;
foreach (Case c in cluster)
{
Case realcase = mybestchoice.Cases.Find(
delegate(Case ca)
{
return(ca.GetFeature("id").GetFeatureValue() == c.GetFeature("id").GetFeatureValue());
}
);
realcase.GetFeature("cluster").SetFeatureValue(t);
c.GetFeature("cluster").SetFeatureValue(t);
for (int j = 0; j < c.GetFeatures().Count; j++)
{
Feature f = (Feature)realcase.GetFeatures()[j];
dG_data.Rows[i].Cells[j].Value = f.GetFeatureValue();
}
i++;
} // end feautres
t++;
} // end clusters
dG_data.AutoSizeColumnsMode = DataGridViewAutoSizeColumnsMode.AllCells;
mybestchoice.Clusters = myclusters;
}
private void button3_Click(object sender, EventArgs e)
{
var watch = System.Diagnostics.Stopwatch.StartNew();
string Standrize_Type = "";
if (rbtn_normalize.Checked)
{
Standrize_Type = "Q";
}
else if (rbtn_minmax.Checked)
{
Standrize_Type = "MN";
}
else
{
Standrize_Type = "none";
}
string Kmeans_Method = "";
if (rbtn_kmeansPlus.Checked)
{
Kmeans_Method = "KPP";
}
else if (rbtn_simpleKmean.Checked)
{
Kmeans_Method = "K";
}
else
{
Kmeans_Method = "none";
}
string Clusters_num = txt_cluster.Text;
mybestchoice = new BestChoice(ArrayCases, Standrize_Type, Kmeans_Method, Clusters_num);
standardize_current_cases = mybestchoice.StandardizeData(ArrayCases);
if (rbtn_simpleKmean.Checked)
{
// Kmeans kmeans = new Kmeans(int.Parse(txt_cluster.Text), standardize_current_cases);
// kmeans.run();
// Number of Attributes
List <Cluster> AllClusters = new List <Cluster>();
AllClusters.Add(null);
Cluster myc = new Cluster();
int Nd = standardize_current_cases[0].GetFeatures().Count - 2;//except id , cluster
Kmeans kmeans = new Kmeans(1, standardize_current_cases);
kmeans.run();
myc.clusters = kmeans.clusters;
foreach (double sk in kmeans.distances)
{
myc.sk += sk;
}
myc.fk = 1;
myc.k = 1;
AllClusters.Add(myc);
for (int ik = 2; ik <= 5; ik++)//19
{
myc = new Cluster();
kmeans = new Kmeans(ik, standardize_current_cases);
kmeans.run();
myc.k = ik;
myc.clusters = kmeans.clusters;
foreach (double sk in kmeans.distances)
{
myc.sk += sk;
}
if (AllClusters[ik - 1].sk == 0)
{
myc.fk = 1;
}
else
{
double ak = 1;
if (Nd > 1)
{
ak = Convert.ToDouble(kmeans.Ak(ik, Nd));
}
myc.fk = myc.sk / (ak * AllClusters[ik - 1].sk);
}
AllClusters.Add(myc);
}
double min = double.MaxValue;
int kk = 0;
for (int ik = 1; ik <= 19; ik++)
{
if (AllClusters[ik].fk < min)
{
kk = AllClusters[ik].k; kmeans.clusters = AllClusters[ik].clusters; min = AllClusters[ik].fk;
}
}
watch.Stop();
textBox1.Text = min.ToString();
txt_cluster.Text = kk.ToString();
var elapsedMs = watch.ElapsedMilliseconds;
label3.Text = "with clustring time " + elapsedMs.ToString();
int t = 0;
int i = 0;
dataGridView1.Rows.Clear();
dataGridView1.Columns.Clear();
for (int j = 0; j < ListCases[0].GetFeatures().Count; j++)
{
Feature feature = (Feature)ListCases[0].GetFeatures()[j];
dataGridView1.Columns.Add(feature.GetFeatureName().ToString(), feature.GetFeatureName().ToString());
}
dataGridView2.Rows.Clear();
dataGridView2.Columns.Clear();
dataGridView1.Rows.Add(ArrayCases.Count);
dataGridView2.Columns.Add("cluster", "cluster");
dataGridView2.Columns.Add("count", "count");
dataGridView2.Rows.Add(kmeans.clusters.Count);
foreach (List <Case> list in kmeans.clusters) // for each cluster, define a new centroid
{
kmeans.centroids[t].GetFeature("id").SetFeatureValue(t);
standardize_current_cases.Add(kmeans.centroids[t]);
dataGridView2.Rows[t].Cells[0].Value = t.ToString();
dataGridView2.Rows[t].Cells[1].Value = list.Count;
mybestchoice.Cost_Function += kmeans.distances[t];
foreach (Case c in list) // Determine the poit on clusters
{
Case realcase = ListCases.Find(
delegate(Case ca)
{
return(ca.GetFeature("id").GetFeatureValue() == c.GetFeature("id").GetFeatureValue());
}
);
realcase.GetFeature("cluster").SetFeatureValue(t);
c.GetFeature("cluster").SetFeatureValue(t);
for (int j = 0; j < c.GetFeatures().Count; j++)
{
Feature f = (Feature)realcase.GetFeatures()[j];
dataGridView1.Rows[i].Cells[j].Value = f.GetFeatureValue();
}
i++;
}
t++;
}
}
else if (rbtn_kmeansPlus.Checked)
{
KmeansPlus mykp = new KmeansPlus();
KmeansPlus.PointClusters myclusters;
List <KmeansPlus.PointClusters> Allmyclusters = new List <KmeansPlus.PointClusters>();
// myclusters = mykp.GetKMeansPP(standardize_current_cases,int.Parse( Clusters_num));
double min_fk = Double.MaxValue;
int min_i = 0;
for (int j = 0; j < 100; j++)
{
myclusters = new KmeansPlus.PointClusters();
myclusters = mykp.GetKMeansPP_K(standardize_current_cases);
Allmyclusters.Add(myclusters);
if (myclusters.Fk < min_fk)
{
min_i = j;
}
}
myclusters = Allmyclusters[min_i];
// KmeansPlus mykp = new KmeansPlus();
// KmeansPlus.PointClusters myclusters =new KmeansPlus.PointClusters();;
// myclusters = mykp.GetKMeansPP_K(standardize_current_cases);
textBox1.Text = myclusters.Fk.ToString();
txt_cluster.Text = myclusters.PC.Count.ToString();
watch.Stop();
var elapsedMs = watch.ElapsedMilliseconds;
label3.Text = "with clustring time " + elapsedMs.ToString();
foreach (Case centroid in myclusters.PC.Keys)
{
standardize_current_cases.Add(centroid);
}
// intialize db grids
int t = 0;
int i = 0;
int countcluster = myclusters.PC.Count;
dataGridView1.Rows.Clear();
dataGridView1.Columns.Clear();
for (int j = 0; j < ListCases[0].GetFeatures().Count; j++)
{
Feature feature = (Feature)ListCases[0].GetFeatures()[j];
dataGridView1.Columns.Add(feature.GetFeatureName().ToString(), feature.GetFeatureName().ToString());
}
dataGridView2.Rows.Clear();
dataGridView2.Columns.Clear();
dataGridView1.Rows.Add(ArrayCases.Count);
dataGridView2.Columns.Add("cluster", "cluster");
dataGridView2.Columns.Add("count", "count");
dataGridView2.Rows.Add(countcluster);
foreach (List <Case> cluster in myclusters.PC.Values)
{
dataGridView2.Rows[t].Cells[0].Value = t.ToString();
dataGridView2.Rows[t].Cells[1].Value = cluster.Count;
foreach (Case c in cluster)
{
Case realcase = ListCases.Find(
delegate(Case ca)
{
return(ca.GetFeature("id").GetFeatureValue() == c.GetFeature("id").GetFeatureValue());
}
);
realcase.GetFeature("cluster").SetFeatureValue(t);
c.GetFeature("cluster").SetFeatureValue(t);
for (int j = 0; j < c.GetFeatures().Count; j++)
{
Feature f = (Feature)realcase.GetFeatures()[j];
dataGridView1.Rows[i].Cells[j].Value = f.GetFeatureValue();
}
i++;
} // end feautres
t++;
} // end clusters
} // end if
else
if (rbtn_nonekmeans.Checked)
{
dataGridView1.DataSource = Db.get_table_contetnt(mybestchoice.TableName);
}
// for sorting
comboBox1.Items.Clear();
for (int i = 0; i < dataGridView1.Columns.Count; i++)
{
comboBox1.Items.Add(dataGridView1.Columns[i].HeaderText);
}
comboBox1.Text = "id";
dataGridView1.AutoSizeColumnsMode =
DataGridViewAutoSizeColumnsMode.AllCells;
}