//groups the top limitOfGroups groups, and compares them with the rest of the dataset.
public List <List <DataSeries> > ConvertToOneVsAllDataSeries(string groupBy, string dim0, string dim1, int limitOfGroups)
{
//need to make sure these guys are linked
var res = new List <List <DataSeries> >();
var schema = new DataFileSchema(this);
schema.ColumnNames.ForEach(col => schema.ColumnNumericMap[col] = false);
schema.ColumnNumericMap[dim0] = true;
schema.ColumnNumericMap[dim1] = true;
schema.GroupBy = groupBy;
var groups = ConvertToDataSeries(schema, limitOfGroups);
var ordered = groups.OrderByDescending(g => g.Rows.Count).ToList();
for (int i = 0; i < Math.Min(10, groups.Count); i++)
{
var temp = new List <DataSeries>();
temp.Add(ordered[i]);
DataSeries Others = new DataSeries(this, schema, ".Others");
for (int j = 0; j < ordered.Count; j++)
{
if (i != j)
{
ordered[j].Rows.ForEach(r => Others.AddRow(r));
}
}
Others.EndInit();
temp.Add(Others);
res.Add(temp);
}
return(res);
}
private void buttonSplatter_Click(object sender, EventArgs e)
{
var dialog = new AddTo1vsAllDialog();
dialog.Text = "Show single splatterplot";
if (listViewDataFiles.SelectedItems.Count > 0)
{
var item = listViewDataFiles.SelectedItems[0];
var dataFile = m_Files[item];
dialog.SetDataFile(dataFile, true);
var res = dialog.ShowDialog(this);
if (res == System.Windows.Forms.DialogResult.OK)
{
var schema = new DataFileSchema(dataFile);
schema.GroupBy = dialog.GroupBy;
var list = dataFile.ConvertToDataSeries(schema, -1);
SplatterModel model = new SplatterModel(list, dialog.HorizontalDim, dialog.VerticalDim, true);
var splatterDialog = new SingleSplatterDialog();
splatterDialog.Text = string.Format("Splatterplot ({0}): {1}", SplatterCount++, dataFile.Name);
splatterDialog.SetModel(model);
splatterDialog.Show();
splatterDialog.BringToFront();
}
}
}
public DataSeries(DataFile dataFile, DataFileSchema schema, string name)
{
m_DataFile = dataFile;
m_Schema = schema;
ColumnData = new Dictionary <string, ColumnData>();
ColumnNames = new List <string>();
Name = name;
}
public DataSeriesRow(DataFileSchema schema, DataRow row)
{
m_Row = row;
foreach (var col in schema.ColumnNames.Where(c => schema.ColumnNumericMap[c]))
{
//m_Values[col] = (float)Math.Log(Convert.ToSingle(row.Field<string>(col)) + 1, 2);
m_Values[col] = Convert.ToSingle(row.Field <string>(col));
}
Selected = false;
}
public List <DataSeries> ConvertToDataSeries(DataFileSchema schema, int limitOfGroups)
{
List <DataSeries> res = new List <DataSeries>();
if (schema.GroupBy.Equals("None"))
{
DataSeries data = new DataSeries(this, schema, Name);
foreach (var row in m_table.AsEnumerable())
{
data.AddRow(new DataSeriesRow(schema, row));
}
data.EndInit();
res.Add(data);
}
else
{
var query =
from row in m_table.AsEnumerable()
group row by row[schema.GroupBy] into set
orderby set.Count() descending
select set;
int max = limitOfGroups > 0 ? limitOfGroups : int.MaxValue;
int index = 0;
DataSeries rest = new DataSeries(this, schema, ".Rest");
foreach (var group in query)
{
if (index < max)
{
DataSeries data = new DataSeries(this, schema, group.Key.ToString());
foreach (var row in group)
{
data.AddRow(new DataSeriesRow(schema, row));
}
data.EndInit();
res.Add(data);
}
else
{
foreach (var row in group)
{
rest.AddRow(new DataSeriesRow(schema, row));
}
}
index++;
}
if (rest.Rows.Count > 0)
{
rest.EndInit();
res.Add(rest);
}
}
return(res);
}
public void SetDataFile(DataFile dataFile)
{
m_Schema = new DataFileSchema(dataFile);
comboBox1.SelectedIndex = 0;
foreach (var name in m_Schema.ColumnNames)
{
AddDialogRow row = new AddDialogRow();
m_Rows.Add(row);
row.SetColumnData(name, m_Schema.ColumnNumericMap[name]);
flowLayoutPanel1.Controls.Add(row);
comboBox1.Items.Add(name);
}
if (m_Rows.Count > 15)
{
m_Rows.ForEach(r => r.SetToIgnore());
}
}
public void AddRowRange(IEnumerable <DataRow> rows, DataFileSchema schema)
{
var list = rows.Select(r => new DataSeriesRow(schema, r));
m_Rows.AddRange(list);
}
