Introduction
Pivot transformation is very useful to summarize data in a flat data table (columns and rows), providing a more clean visualization of the data.
In this article, we will see two methods to retrieve Pivot and Inverted tables from a
DataTable. These methods are very useful, specially for those using metadata to retrieve information from a database, or for those who can not use Pivot transformation from SQL Server 2005.
The two methods are independent, but complement each other in a way that data can be manipulated to return the table inverted or a more complexpivot table.
The Pivot Methods
First - Simple Inversion
Read all data, and return all columns as lines and lines as columns.
The X axis column provided is used as the column header, and some columns may be ignored in the process, if desired so.
/// <summary>
/// Gets a Inverted DataTable
/// </summary>
/// <param name="table">DataTable do invert</param>
/// <param name="columnX">X Axis Column</param>
/// <param name="nullValue">null Value to Complete the Pivot Table</param>
/// <param name="columnsToIgnore">Columns that should be ignored in the pivot
/// process (X Axis column is ignored by default)</param>
/// <returns>C# Pivot Table Method - Felipe Sabino</returns>
public static DataTable GetInversedDataTable(DataTable table, string columnX,
params string[] columnsToIgnore)
{
//Create a DataTable to Return
DataTable returnTable = new DataTable();
if (columnX == "")
columnX = table.Columns[0].ColumnName;
//Add a Column at the beginning of the table
returnTable.Columns.Add(columnX);
//Read all DISTINCT values from columnX Column in the provided DataTale
List<string> columnXValues = new List<string>();
//Creates list of columns to ignore
List<string> listColumnsToIgnore = new List<string>();
if (columnsToIgnore.Length > 0)
listColumnsToIgnore.AddRange(columnsToIgnore);
if (!listColumnsToIgnore.Contains(columnX))
listColumnsToIgnore.Add(columnX);
foreach (DataRow dr in table.Rows)
{
string columnXTemp = dr[columnX].ToString();
//Verify if the value was already listed
if (!columnXValues.Contains(columnXTemp))
{
//if the value id different from others provided, add to the list of
//values and creates a new Column with its value.
columnXValues.Add(columnXTemp);
returnTable.Columns.Add(columnXTemp);
}
else
{
//Throw exception for a repeated value
throw new Exception("The inversion used must have " +
"unique values for column " + columnX);
}
}
//Add a line for each column of the DataTable
foreach (DataColumn dc in table.Columns)
{
if (!columnXValues.Contains(dc.ColumnName) &&
!listColumnsToIgnore.Contains(dc.ColumnName))
{
DataRow dr = returnTable.NewRow();
dr[0] = dc.ColumnName;
returnTable.Rows.Add(dr);
}
}
//Complete the datatable with the values
for (int i = 0; i < returnTable.Rows.Count; i++)
{
for (int j = 1; j < returnTable.Columns.Count; j++)
{
returnTable.Rows[i][j] =
table.Rows[j - 1][returnTable.Rows[i][0].ToString()].ToString();
}
}
return returnTable;
}
Second - Pivoting
It uses the idea of three axes to build the new table. The X axis column is the column with the Unique Values to build the columns header. The Y axis value is the column with the values to be displayed as the rows in the first column. The Z axis is the "value" and is the match of X and Y in the DataSource, and can be the sum of values if more than one different value is found in the process. The null value is provided in case there is the need to fill empty cells of the table with a certain value.
The flag to sum values is used in case there is more than one value for a certain X and Y column combination; if it is "false", the last value that is read is displayed.
/// <summary>
/// Gets a Inverted DataTable
/// </summary>
/// <param name="table">Provided DataTable</param>
/// <param name="columnX">X Axis Column</param>
/// <param name="columnY">Y Axis Column</param>
/// <param name="columnZ">Z Axis Column (values)</param>
/// <param name="columnsToIgnore">Whether to ignore some column, it must be
/// provided here</param>
/// <param name="nullValue">null Values to be filled</param>
/// <returns>C# Pivot Table Method - Felipe Sabino</returns>
public static DataTable GetInversedDataTable(DataTable table, string columnX,
string columnY, string columnZ, string nullValue, bool sumValues)
{
//Create a DataTable to Return
DataTable returnTable = new DataTable();
if (columnX == "")
columnX = table.Columns[0].ColumnName;
//Add a Column at the beginning of the table
returnTable.Columns.Add(columnY);
//Read all DISTINCT values from columnX Column in the provided DataTale
List<string> columnXValues = new List<string>();
foreach (DataRow dr in table.Rows)
{
string columnXTemp = dr[columnX].ToString();
if (!columnXValues.Contains(columnXTemp))
{
//Read each row value, if it's different from others provided, add to
//the list of values and creates a new Column with its value.
columnXValues.Add(columnXTemp);
returnTable.Columns.Add(columnXTemp);
}
}
//Verify if Y and Z Axis columns re provided
if (columnY != "" && columnZ != "")
{
//Read DISTINCT Values for Y Axis Column
List<string> columnYValues = new List<string>();
foreach (DataRow dr in table.Rows)
{
if (!columnYValues.Contains(dr[columnY].ToString()))
columnYValues.Add(dr[columnY].ToString());
}
//Loop all Column Y Distinct Value
foreach (string columnYValue in columnYValues)
{
//Creates a new Row
DataRow drReturn = returnTable.NewRow();
drReturn[0] = columnYValue;
//foreach column Y value, The rows are selected distincted
DataRow[] rows = table.Select(columnY + "='" + columnYValue + "'");
//Read each row to fill the DataTable
foreach (DataRow dr in rows)
{
string rowColumnTitle = dr[columnX].ToString();
//Read each column to fill the DataTable
foreach (DataColumn dc in returnTable.Columns)
{
if (dc.ColumnName == rowColumnTitle)
{
//If Sum of Values is True it try to perform a Sum
//If sum is not possible due to value types, the value
// displayed is the last one read
if (sumValues)
{
try
{
drReturn[rowColumnTitle] =
Convert.ToDecimal(drReturn[rowColumnTitle]) +
Convert.ToDecimal(dr[columnZ]);
}
catch
{
drReturn[rowColumnTitle] = dr[columnZ];
}
}
else
{
drReturn[rowColumnTitle] = dr[columnZ];
}
}
}
}
returnTable.Rows.Add(drReturn);
}
}
else
{
throw new Exception("The columns to perform inversion are not provided");
}
//if a nullValue is provided, fill the datable with it
if (nullValue != "")
{
foreach (DataRow dr in returnTable.Rows)
{
foreach (DataColumn dc in returnTable.Columns)
{
if (dr[dc.ColumnName].ToString() == "")
dr[dc.ColumnName] = nullValue;
}
}
}
return returnTable;
}
Using the Code
The article provides two ways to perform a Pivot transformation.
In both ways, the table below will be used as an example of the data source:
| EmployeeID | OrderID | Amount | Cost | Date |
| Sam | 1 | 25 | 13 | 01/10/2007 |
| Sam | 2 | 512 | 1 | 02/10/2007 |
| Sam | 3 | 512 | 1 | 03/10/2007 |
| Tom | 4 | 50 | 1 | 04/10/2007 |
| Tom | 5 | 3 | 7 | 03/10/2007 |
| Tom | 6 | 78,75 | 12 | 02/10/2007 |
| Sue | 7 | 11 | 7 | 01/10/2007 |
| Sue | 8 | 2,5 | 66,2 | 02/10/2007 |
| Sue | 9 | 2,5 | 22 | 03/10/2007 |
| Jack | 10 | 6 | 23 | 02/10/2007 |
| Jack | 11 | 117 | 199 | 04/10/2007 |
| Jack | 12 | 13 | 2,6 | 01/10/2007 |
| Jack | 13 | 11,4 | 99,8 | 03/10/2007 |
| Phill | 14 | 37 | 2,1 | 02/10/2007 |
| Phill | 15 | 65,2 | 99,3 | 04/10/2007 |
| Phill | 16 | 34,1 | 27 | 02/10/2007 |
| Phill | 17 | 17 | 959 | 04/10/2007 |
The code below shows how to create the table above:
DataTable dt = new DataTable();
dt.Columns.Add("EmployeeID", Type.GetType("System.String"));
dt.Columns.Add("OrderID", Type.GetType("System.Int32"));
dt.Columns.Add("Amount", Type.GetType("System.Decimal"));
dt.Columns.Add("Cost", Type.GetType("System.Decimal"));
dt.Columns.Add("Date", Type.GetType("System.String"));
dt.Rows.Add(new object[] { "Sam", 1, 25.00, 13.00, "01/10/2007" });
dt.Rows.Add(new object[] { "Sam", 2, 512.00, 1.00, "02/10/2007" });
dt.Rows.Add(new object[] { "Sam", 3, 512.00, 1.00, "03/10/2007" });
dt.Rows.Add(new object[] { "Tom", 4, 50.00, 1.00, "04/10/2007" });
dt.Rows.Add(new object[] { "Tom", 5, 3.00, 7.00, "03/10/2007" });
dt.Rows.Add(new object[] { "Tom", 6, 78.75, 12.00, "02/10/2007" });
dt.Rows.Add(new object[] { "Sue", 7, 11.00, 7.00, "01/10/2007" });
dt.Rows.Add(new object[] { "Sue", 8, 2.50, 66.20, "02/10/2007" });
dt.Rows.Add(new object[] { "Sue", 9, 2.50, 22.00, "03/10/2007" });
dt.Rows.Add(new object[] { "Jack", 10, 6.00, 23.00, "02/10/2007" });
dt.Rows.Add(new object[] { "Jack", 11, 117.00, 199.00, "04/10/2007" });
dt.Rows.Add(new object[] { "Jack", 12, 13.00, 2.60, "01/10/2007" });
dt.Rows.Add(new object[] { "Jack", 13, 11.40, 99.80, "03/10/2007" });
dt.Rows.Add(new object[] { "Phill", 14, 37.00, 2.10, "02/10/2007" });
dt.Rows.Add(new object[] { "Phill", 15, 65.20, 99.30, "04/10/2007" });
dt.Rows.Add(new object[] { "Phill", 16, 34.10, 27.00, "02/10/2007" });
dt.Rows.Add(new object[] { "Phill", 17, 17.00, 959.00, "04/10/2007" });
First - Simple Inversion
A column is provided and the
DataTable has its "axes turned". This method makes it possible to ignore some columns during the inversion, so that the return table has only the columns of interest. If we want to invert with the column "OrderID" as the title (X axis column), we should use the code below://For example, for the DataTable provided as Datasource
DataTable dtReturn = GetInversedDataTable(dt, "OrderID");
The return table will be like the one below:
Second - Pivot Table
The three columns are provided and a new
DataTable is returned.
The example below will use the source table and the params below to build a Pivot Table.
- X axis column: "Date"
- Y axis column: "EmployeeID"
- Z axis column: "Cost"
- Null value: "-";
- Sum of values: true
//For example, for the DataTable provided as Datasource
DataTable dtReturn = GetInversedDataTable(dt, "Date", "EmployeeID",
"Cost", "-", true);
The return table will be like the one below:
| EmployeeID | 01/12/2007 | 02/12/2007 | 03/12/2007 | 04/12/2007 |
| Sam | 13 | 1 | 1 | - |
| Tom | - | 12 | 7 | 1 |
| Sue | 7 | 66,2 | 22 | - |
| Jack | 2,6 | 23 | 99,8 | 199 |
| Phill | - | 27 | - | 959 |
Points of Interest
You can read more about Pivot Tables at:
- http://en.wikipedia.org/wiki/Pivot_table
- http://books.slashdot.org/article.pl?sid=04/11/19/2012256
- http://j-walk.com/ss/excel/usertips/tip068.htm
History
- 10/12/2007 - First release.
License
This article, along with any associated source code and files, is licensed under The Code Project Open License (CPOL)
