/// <summary>
/// Ensures that the selected columns in the source <see cref="DataColumnCollection"/> do also exist in the destination <see cref="DataColumnCollection"/>.
/// </summary>
/// <param name="sourceTable">The source table.</param>
/// <param name="selectedSourceDataColumns">The selected source data columns. If this parameter is null, all source columns are selected.</param>
/// <param name="destinationTable">The destination table.</param>
public static void EnsureColumnsExistInDestinationCollection(this DataColumnCollection sourceTable, IAscendingIntegerCollection selectedSourceDataColumns, DataColumnCollection destinationTable)
{
if (null == selectedSourceDataColumns || 0 == selectedSourceDataColumns.Count)
{
selectedSourceDataColumns = Altaxo.Collections.ContiguousIntegerRange.FromStartAndCount(0, sourceTable.ColumnCount);
}
foreach (var colIdx in selectedSourceDataColumns)
{
var srcCol = sourceTable[colIdx];
destinationTable.EnsureExistence(
sourceTable.GetColumnName(srcCol),
srcCol.GetType(),
sourceTable.GetColumnKind(srcCol),
sourceTable.GetColumnGroup(srcCol)
);
}
}
/// <summary>
/// Calculates statistics of selected columns. Creates a new table where the statistical data will be written to.
/// </summary>
/// <param name="srctable">Source table.</param>
/// <param name="selectedColumns">Selected data columns in the source table. If the argument is null, all columns will be used.</param>
/// <param name="selectedRows">Selected rows in the source table. If the argument is null, all rows will be used.</param>
/// <param name="destinationTable">The table where the statistical results are written to.</param>
public static void DoStatisticsOnColumns(
this DataColumnCollection srctable,
IAscendingIntegerCollection selectedColumns,
IAscendingIntegerCollection selectedRows,
DataColumnCollection destinationTable
)
{
bool bUseSelectedColumns = (null != selectedColumns && 0 != selectedColumns.Count);
int numcols = bUseSelectedColumns ? selectedColumns.Count : srctable.ColumnCount;
bool bUseSelectedRows = (null != selectedRows && 0 != selectedRows.Count);
if (numcols == 0)
{
return; // nothing selected
}
// add a text column and some double columns
// note: statistics is only possible for numeric columns since
// otherwise in one column doubles and i.e. dates are mixed, which is not possible
// 1st column is the name of the column of which the statistics is made
var colCol = new Data.TextColumn();
// 2nd column is the mean
var colMean = new Data.DoubleColumn();
// 3rd column is the standard deviation
var colSd = new Data.DoubleColumn();
// 4th column is the standard e (N)
var colSe = new Data.DoubleColumn();
// 5th column is the sum
var colSum = new Data.DoubleColumn();
var colSumSqr = new Data.DoubleColumn();
// 6th column is the number of items for statistics
var colN = new Data.DoubleColumn();
var colFracOneSigma = new Data.DoubleColumn();
var colFracTwoSigma = new Data.DoubleColumn();
var colFracThreeSigma = new Data.DoubleColumn();
var colMinimum = new DoubleColumn(); // Minimum of the values
var colMaximum = new DoubleColumn(); // Maximum of the values
int currRow = 0;
for (int si = 0; si < numcols; si++)
{
Altaxo.Data.DataColumn col = bUseSelectedColumns ? srctable[selectedColumns[si]] : srctable[si];
if (!(col is Altaxo.Data.INumericColumn))
{
continue;
}
int rows = bUseSelectedRows ? selectedRows.Count : srctable.RowCount;
if (rows == 0)
{
continue;
}
// now do the statistics
var ncol = (Data.INumericColumn)col;
double sum = 0;
double sumsqr = 0;
int NN = 0;
double minimum = double.PositiveInfinity;
double maximum = double.NegativeInfinity;
for (int i = 0; i < rows; i++)
{
double val = bUseSelectedRows ? ncol[selectedRows[i]] : ncol[i];
if (double.IsNaN(val))
{
continue;
}
NN++;
sum += val;
sumsqr += (val * val);
minimum = Math.Min(minimum, val);
maximum = Math.Max(maximum, val);
}
// now fill a new row in the worksheet
double mean = sum / NN;
double ymy0sqr = sumsqr - sum * sum / NN;
if (ymy0sqr < 0)
{
ymy0sqr = 0; // if this is lesser zero, it is a rounding error, so set it to zero
}
double sd = NN > 1 ? Math.Sqrt(ymy0sqr / (NN - 1)) : 0;
double se = sd / Math.Sqrt(NN);
// calculate fractions
double oneSigmaLo = mean - 1 * sd, oneSigmaHi = mean + 1 * sd;
double twoSigmaLo = mean - 2 * sd, twoSigmaHi = mean + 2 * sd;
double threeSigmaLo = mean - 3 * sd, threeSigmaHi = mean + 3 * sd;
int cntOneSigma = 0, cntTwoSigma = 0, cntThreeSigma = 0;
for (int i = 0; i < rows; i++)
{
double val = bUseSelectedRows ? ncol[selectedRows[i]] : ncol[i];
if (double.IsNaN(val))
{
continue;
}
if (Altaxo.Calc.RMath.IsInIntervalCC(val, oneSigmaLo, oneSigmaHi))
{
++cntOneSigma;
}
if (Altaxo.Calc.RMath.IsInIntervalCC(val, twoSigmaLo, twoSigmaHi))
{
++cntTwoSigma;
}
if (Altaxo.Calc.RMath.IsInIntervalCC(val, threeSigmaLo, threeSigmaHi))
{
++cntThreeSigma;
}
}
if (0 == NN)
{
minimum = maximum = double.NaN;
}
colCol[currRow] = col.Name;
colMean[currRow] = mean; // mean
colSd[currRow] = sd;
colSe[currRow] = se;
colSum[currRow] = sum;
colSumSqr[currRow] = sumsqr;
colN[currRow] = NN;
colFracOneSigma[currRow] = cntOneSigma / (double)NN;
colFracTwoSigma[currRow] = cntTwoSigma / (double)NN;
colFracThreeSigma[currRow] = cntThreeSigma / (double)NN;
colMinimum[currRow] = minimum;
colMaximum[currRow] = maximum;
currRow++; // for the next column
} // for all selected columns
if (currRow != 0)
{
destinationTable.EnsureExistence(DefaultColumnNameColumnName, typeof(TextColumn), ColumnKind.X, 0).Append(colCol);
AppendStatisticalData(destinationTable, colMean, colSd, colSe, colSum, colSumSqr, colN, colFracOneSigma, colFracTwoSigma, colFracThreeSigma, colMinimum, colMaximum);
}
}
private static void AppendStatisticalData(DataColumnCollection destinationTable, Data.DoubleColumn colMean, Data.DoubleColumn colSd, Data.DoubleColumn colSe, Data.DoubleColumn colSum, Data.DoubleColumn colSumSqr, Data.DoubleColumn colN, Data.DoubleColumn fracOneSigma, Data.DoubleColumn fracTwoSigma, Data.DoubleColumn fracThreeSigma, DoubleColumn minimum, DoubleColumn maximum)
{
destinationTable.EnsureExistence(DefaultMeanColumnName, typeof(DoubleColumn), ColumnKind.V, 0).Append(colMean);
destinationTable.EnsureExistence(DefaultStandardErrorColumnName, typeof(DoubleColumn), ColumnKind.Err, 0).Append(colSe);
destinationTable.EnsureExistence(DefaultStandardDeviationColumnName, typeof(DoubleColumn), ColumnKind.V, 0).Append(colSd);
destinationTable.EnsureExistence(DefaultSumColumnName, typeof(DoubleColumn), ColumnKind.V, 0).Append(colSum);
destinationTable.EnsureExistence(DefaultSumSqrColumnName, typeof(DoubleColumn), ColumnKind.V, 0).Append(colSumSqr);
destinationTable.EnsureExistence(DefaultNumberOfItemsColumnName, typeof(DoubleColumn), ColumnKind.V, 0).Append(colN);
destinationTable.EnsureExistence(DefaultFractionInOneSigmaColumnName, typeof(DoubleColumn), ColumnKind.V, 0).Append(fracOneSigma);
destinationTable.EnsureExistence(DefaultFractionInTwoSigmaColumnName, typeof(DoubleColumn), ColumnKind.V, 0).Append(fracTwoSigma);
destinationTable.EnsureExistence(DefaultFractionInThreeSigmaColumnName, typeof(DoubleColumn), ColumnKind.V, 0).Append(fracThreeSigma);
destinationTable.EnsureExistence(DefaultMinimumColumnName, typeof(DoubleColumn), ColumnKind.V, 0).Append(minimum);
destinationTable.EnsureExistence(DefaultMaximumColumnName, typeof(DoubleColumn), ColumnKind.V, 0).Append(maximum);
}
/// <summary>
/// Calculates statistics of selected columns. Creates a new table where the statistical data will be written to.
/// </summary>
/// <param name="srctable">Source table.</param>
/// <param name="selectedColumns">Selected data columns in the source table.</param>
/// <param name="selectedRows">Selected rows in the source table.</param>
/// <param name="destinationTable">The table where the statistical results are written to.</param>
public static void DoStatisticsOnRows(
this DataColumnCollection srctable,
IAscendingIntegerCollection selectedColumns,
IAscendingIntegerCollection selectedRows,
DataColumnCollection destinationTable
)
{
bool bUseSelectedColumns = (null != selectedColumns && 0 != selectedColumns.Count);
int numcols = bUseSelectedColumns ? selectedColumns.Count : srctable.ColumnCount;
if (numcols == 0)
{
return; // nothing selected
}
bool bUseSelectedRows = (null != selectedRows && 0 != selectedRows.Count);
int numrows = bUseSelectedRows ? selectedRows.Count : srctable.RowCount;
if (numrows == 0)
{
return;
}
var cRows = new DoubleColumn();
// 1st column is the mean, and holds the sum during the calculation
var colMean = new Data.DoubleColumn();
// 2rd column is the standard deviation, and holds the square sum during calculation
var colSD = new Data.DoubleColumn();
// 3th column is the standard e (N)
var colSE = new Data.DoubleColumn();
// 4th column is the sum
var colSum = new Data.DoubleColumn();
// 5th column is the number of items for statistics
var colNN = new Data.DoubleColumn();
var colSumSqr = new Data.DoubleColumn();
var colFracOneSigma = new Data.DoubleColumn();
var colFracTwoSigma = new Data.DoubleColumn();
var colFracThreeSigma = new Data.DoubleColumn();
var colMinimum = new DoubleColumn();
var colMaximum = new DoubleColumn();
// first fill the cols c1, c2, c5 with zeros because we want to sum up
for (int i = 0; i < numrows; i++)
{
colSum[i] = 0;
colSumSqr[i] = 0;
colNN[i] = 0;
colMinimum[i] = double.PositiveInfinity;
colMaximum[i] = double.NegativeInfinity;
}
for (int si = 0; si < numcols; si++)
{
Altaxo.Data.DataColumn col = bUseSelectedColumns ? srctable[selectedColumns[si]] : srctable[si];
if (!(col is Altaxo.Data.INumericColumn))
{
continue;
}
// now do the statistics
var ncol = (Data.INumericColumn)col;
for (int i = 0; i < numrows; i++)
{
int row = bUseSelectedRows ? selectedRows[i] : i;
cRows[i] = row;
double val = ncol[row];
if (double.IsNaN(val))
{
continue;
}
colSum[i] += val;
colSumSqr[i] += val * val;
colNN[i] += 1;
colMinimum[i] = Math.Min(colMinimum[i], val);
colMaximum[i] = Math.Max(colMaximum[i], val);
}
} // for all selected columns
// now calculate the statistics
for (int i = 0; i < numrows; i++)
{
// now fill a new row in the worksheet
double NN = colNN[i];
double sum = colSum[i];
double sumsqr = colSumSqr[i];
if (NN > 0)
{
double mean = sum / NN;
double ymy0sqr = sumsqr - sum * sum / NN;
if (ymy0sqr < 0)
{
ymy0sqr = 0; // if this is lesser zero, it is a rounding error, so set it to zero
}
double sd = NN > 1 ? Math.Sqrt(ymy0sqr / (NN - 1)) : 0;
double se = sd / Math.Sqrt(NN);
colMean[i] = mean; // mean
colSD[i] = sd;
colSE[i] = se;
}
else
{
colMinimum[i] = double.NaN;
colMaximum[i] = double.NaN;
}
} // for all rows
// calculate fractions
for (int i = 0; i < numrows; i++)
{
int row = bUseSelectedRows ? selectedRows[i] : i;
double mean = colMean[i];
double sd = colSD[i];
// calculate fractions
double oneSigmaLo = mean - 1 * sd, oneSigmaHi = mean + 1 * sd;
double twoSigmaLo = mean - 2 * sd, twoSigmaHi = mean + 2 * sd;
double threeSigmaLo = mean - 3 * sd, threeSigmaHi = mean + 3 * sd;
int cntOneSigma = 0, cntTwoSigma = 0, cntThreeSigma = 0;
for (int si = 0; si < numcols; si++)
{
Altaxo.Data.DataColumn col = bUseSelectedColumns ? srctable[selectedColumns[si]] : srctable[si];
if (!(col is Altaxo.Data.INumericColumn))
{
continue;
}
// now do the statistics
var ncol = (Data.INumericColumn)col;
double val = ncol[row];
if (double.IsNaN(val))
{
continue;
}
if (Altaxo.Calc.RMath.IsInIntervalCC(val, oneSigmaLo, oneSigmaHi))
{
++cntOneSigma;
}
if (Altaxo.Calc.RMath.IsInIntervalCC(val, twoSigmaLo, twoSigmaHi))
{
++cntTwoSigma;
}
if (Altaxo.Calc.RMath.IsInIntervalCC(val, threeSigmaLo, threeSigmaHi))
{
++cntThreeSigma;
}
}
colFracOneSigma[i] = cntOneSigma / colNN[i];
colFracTwoSigma[i] = cntTwoSigma / colNN[i];
colFracThreeSigma[i] = cntThreeSigma / colNN[i];
}
destinationTable.EnsureExistence(DefaultRowNumberColumnName, typeof(DoubleColumn), ColumnKind.X, 0).Append(cRows);
AppendStatisticalData(destinationTable, colMean, colSD, colSE, colSum, colSumSqr, colNN, colFracOneSigma, colFracTwoSigma, colFracThreeSigma, colMinimum, colMaximum);
}
public static DataColumnCollection GetPropertyColumns(this Origin.Worksheet wks)
{
if (null == wks)
{
throw new ArgumentNullException("wks");
}
var result = new DataColumnCollection();
// I found no way to ask, if a label column is used or not
// therefore, we have to try all cells inside the longname, the units and the comments label column
var labelCols = new Dictionary <string, Altaxo.Data.TextColumn>();
DataColumn destLongNameCol = null, destUnitCol = null, destCommentCol = null;
var paraCol = new DataColumn[20];
var srcDataCols = wks.Cols;
for (int i = 0; i < srcDataCols; ++i)
{
var srcCol = wks.Columns[i];
if (!string.IsNullOrEmpty(srcCol.LongName))
{
if (null == destLongNameCol)
{
destLongNameCol = result.EnsureExistence("LongName", typeof(TextColumn), ColumnKind.V, 0);
}
destLongNameCol[i] = srcCol.LongName;
}
if (!string.IsNullOrEmpty(srcCol.Units))
{
if (null == destUnitCol)
{
destUnitCol = result.EnsureExistence("Unit", typeof(TextColumn), ColumnKind.V, 0);
}
destUnitCol[i] = srcCol.Units;
}
if (!string.IsNullOrEmpty(srcCol.Comments))
{
if (null == destCommentCol)
{
destCommentCol = result.EnsureExistence("Comments", typeof(TextColumn), ColumnKind.V, 0);
}
destCommentCol[i] = srcCol.Comments;
}
for (int nPara = 0; nPara <= 11; ++nPara)
{
if (!string.IsNullOrEmpty(srcCol.Parameter[nPara]))
{
if (null == paraCol[nPara])
{
paraCol[nPara] = result.EnsureExistence("Parameter" + nPara.ToString(), typeof(TextColumn), ColumnKind.V, 0);
}
paraCol[nPara][i] = srcCol.Parameter[nPara];
}
}
}
return(result);
}
