public void SetFilter(bool[] remainingRows)
{
filteredData = (IPreprocessingData)originalData.Clone();
filteredData.InTransaction(() => {
var remainingIndices = Enumerable.Range(0, remainingRows.Length).Where(x => remainingRows[x]);
foreach (var v in filteredData.VariableNames)
{
var ci = filteredData.GetColumnIndex(v);
if (filteredData.VariableHasType <double>(ci))
{
var values = filteredData.GetValues <double>(ci);
var filteredValues = remainingIndices.Select(x => values[x]).ToList();
filteredData.SetValues(ci, filteredValues);
}
else if (filteredData.VariableHasType <DateTime>(ci))
{
var values = filteredData.GetValues <DateTime>(ci);
var filteredValues = remainingIndices.Select(x => values[x]).ToList();
filteredData.SetValues(ci, filteredValues);
}
else if (filteredData.VariableHasType <string>(ci))
{
var values = filteredData.GetValues <string>(ci);
var filteredValues = remainingIndices.Select(x => values[x]).ToList();
filteredData.SetValues(ci, filteredValues);
}
}
});
OnFilterChanged();
}
public void PersistFilter()
{
originalData.InTransaction(() => {
for (int i = 0; i < filteredData.Columns; ++i)
{
if (filteredData.VariableHasType <double>(i))
{
originalData.SetValues <double>(i, filteredData.GetValues <double>(i));
}
else if (filteredData.VariableHasType <string>(i))
{
originalData.SetValues <string>(i, filteredData.GetValues <string>(i));
}
else if (filteredData.VariableHasType <DateTime>(i))
{
originalData.SetValues <DateTime>(i, filteredData.GetValues <DateTime>(i));
}
else
{
throw new ArgumentException("Data types of columns do not match");
}
}
});
ResetFilter();
}
public static IEnumerable <string> GetVariableNamesForGrouping(IPreprocessingData preprocessingData, int maxDistinctValues = 20)
{
var variableNames = new List <string>();
for (int i = 0; i < preprocessingData.Columns; ++i)
{
int distinctValues = Int32.MaxValue;
if (preprocessingData.VariableHasType <double>(i))
{
distinctValues = preprocessingData.GetValues <double>(i).GroupBy(x => x).Count();
}
else if (preprocessingData.VariableHasType <string>(i))
{
distinctValues = preprocessingData.GetValues <string>(i).GroupBy(x => x).Count();
}
else if (preprocessingData.VariableHasType <DateTime>(i))
{
distinctValues = preprocessingData.GetValues <DateTime>(i).GroupBy(x => x).Count();
}
if (distinctValues <= maxDistinctValues)
{
variableNames.Add(preprocessingData.GetVariableName(i));
}
}
return(variableNames);
}
private void PreserveColumns(IEnumerable <Transformation <double> > transformations)
{
foreach (var transformation in transformations)
{
if (!originalColumns.ContainsKey(transformation.Column))
{
int colIndex = preprocessingData.GetColumnIndex(transformation.Column);
var originalData = preprocessingData.GetValues <double>(colIndex);
originalColumns.Add(transformation.Column, originalData);
}
}
}
public IList <T> GetValues <T>(int columnIndex, bool considerSelection)
{
return(ActiveData.GetValues <T>(columnIndex, considerSelection));
}