public IntermediateColumn(string name, DataViewType type, ColumnPurpose purpose, ColumnDimensions dimensions)
{
ColumnName = name;
Type = type;
Purpose = purpose;
Dimensions = dimensions;
}
public GroupingColumn(string name, DataKind kind, ColumnPurpose purpose, string rangeSelector)
{
SuggestedName = name;
ItemKind = kind;
Purpose = purpose;
ColumnRangeSelector = rangeSelector;
}
public GroupingColumn(string name, DataKind kind, ColumnPurpose purpose, Range[] ranges)
{
SuggestedName = name;
ItemKind = kind;
Purpose = purpose;
Ranges = ranges;
}
public DatasetColumnInfo(string name, DataViewType type, ColumnPurpose purpose, ColumnDimensions dimensions)
{
Name = name;
Type = type;
Purpose = purpose;
Dimensions = dimensions;
}
public InferredColumn(string name, DataKind type, TextLoader.Range[] source, ColumnPurpose columnPurpose)
{
Name = name;
Type = type;
Source = source;
ColumnPurpose = columnPurpose;
}
public IntermediateColumn(IDataView data, int columnId, ColumnPurpose suggestedPurpose = ColumnPurpose.Ignore)
{
_data = data;
_columnId = columnId;
_type = new Lazy <DataViewType>(() => _data.Schema[_columnId].Type);
_columnName = new Lazy <string>(() => _data.Schema[_columnId].Name);
_suggestedPurpose = suggestedPurpose;
}
private static int GetPurposeGroupId(int columnIndex, ColumnPurpose purpose)
{
if (purpose == ColumnPurpose.CategoricalFeature || purpose == ColumnPurpose.TextFeature)
{
return(columnIndex);
}
return(0);
}
public virtual bool IgnoreColumn(ColumnPurpose purpose)
{
if (purpose != ColumnPurpose.TextFeature &&
purpose != ColumnPurpose.CategoricalFeature &&
purpose != ColumnPurpose.NumericFeature)
{
return(true);
}
return(false);
}
public Column(string name, ColumnPurpose purpose, DataKind?dataKind, string ranges)
{
Contracts.CheckValue(name, nameof(name));
Contracts.CheckValue(ranges, nameof(ranges));
_name = name;
ColumnPurpose = purpose;
Kind = dataKind;
Indices = ColumnGroupingInference.GetRange(ranges);
}
private static string GetName(DataKind itemKind, ColumnPurpose purpose, List <GroupingColumn> previousColumns)
{
string prefix = GetPurposeName(purpose, itemKind);
int i = 0;
string name = prefix;
while (previousColumns.Any(x => x.SuggestedName == name))
{
i++;
name = string.Format("{0}{1:00}", prefix, i);
}
return(name);
}
public IntermediateColumn(IDataView data, PurposeInference.Column column)
{
_data = data;
_columnId = column.ColumnIndex;
_purpose = column.Purpose;
_type = new Lazy <ColumnType>(() => _data.Schema[_columnId].Type);
_columnName = new Lazy <string>(() => _data.Schema[_columnId].Name);
_hasMissing = new Lazy <bool>(() =>
{
if (Type.ItemType() != NumberType.R4)
{
return(false);
}
return(Type.IsVector() ? HasMissingVector() : HasMissingOne());
});
}
private static string GetPurposeName(ColumnPurpose purpose, DataKind itemKind)
{
switch (purpose)
{
case ColumnPurpose.NumericFeature:
if (itemKind == DataKind.Bool)
{
return("BooleanFeatures");
}
else
{
return("Features");
}
case ColumnPurpose.CategoricalFeature:
return("Cat");
default:
return(Enum.GetName(typeof(ColumnPurpose), purpose));
}
}
public Column(int columnIndex, ColumnPurpose purpose, DataKind itemKind)
{
ColumnIndex = columnIndex;
Purpose = purpose;
ItemKind = itemKind;
}
public Column(int columnIndex, ColumnPurpose purpose)
{
ColumnIndex = columnIndex;
Purpose = purpose;
}
public Column(string name, ColumnType columnType, ColumnPurpose columnPurpose)
{
this.Name = name;
this.ColumnPurpose = columnPurpose;
this.ColumnType = columnType;
}
public override bool IgnoreColumn(ColumnPurpose purpose)
{
return(purpose != ColumnPurpose.NumericFeature);
}
