How to convert DataTable columns to IEnumerable[] which is required to create a data frame in R.NET
I have the following code:
DataTable dt = CreateDateTable();
REngine e = REngine.GetInstance();
IEnumerable[] columns = new IEnumerable[dt.Columns.Count];
string[] columnNames = dt.Columns.Cast<DataColumn>()
.Select(x => x.ColumnName)
.ToArray();
for(int i=0; i<dt.Columns.Count; i++)
//This is the place where I am stuck. How to convert column to base type array instead of object array
columns[i] = dt.Rows.Cast<DataRow>().Select(row => row[i]).ToArray();
DataFrame df = e.CreateDataFrame(columns: columns,
columnNames: columnNames,
stringsAsFactors: false);
I get the following exception:
Test 'XXX.ReadResultsTest' failed: System.NotSupportedException : Cannot convert type System.Object[] to an R vector
w RDotNet.REngineExtension.ToVector(REngine engine, IEnumerable values)
w System.Array.ConvertAll[TInput,TOutput](TInput[] array, Converter`2 converter)
w RDotNet.REngineExtension.CreateDataFrame(REngine engine, IEnumerable[] columns, String[] columnNames, String[] rowNames, Boolean checkRows, Boolean checkNames, Boolean stringsAsFactors)
DataTable
has columns of different types and I don't know what are the types so I cannot do like in this example with double
:
for (int i = 0; i < dt.Columns.Count; i++)
columns[i] = dt.Rows.Cast<DataRow>().Select(row => row.Field<double>(i)).ToArray();
UPDATE
So far I have an ugly solution, can it be done better?
for (int i = 0; i < dt.Columns.Count; i++)
{
switch (Type.GetTypeCode(dt.Columns[i].DataType))
{
case TypeCode.String:
columns[i] = dt.Rows.Cast<DataRow>().Select(row => row.Field<string>(i)).ToArray();
break;
case TypeCode.Double:
columns[i] = dt.Rows.Cast<DataRow>().Select(row => row.Field<double>(i)).ToArray();
break;
case TypeCode.Int32:
columns[i] = dt.Rows.Cast<DataRow>().Select(row => row.Field<int>(i)).ToArray();
break;
case TypeCode.Int64:
columns[i] = dt.Rows.Cast<DataRow>().Select(row => row.Field<long>(i)).ToArray();
break;
default:
//columns[i] = dt.Rows.Cast<DataRow>().Select(row => row[i]).ToArray();
throw new InvalidOperationException(String.Format("Type {0} is not supported", dt.Columns[i].DataType.Name));
}
}