dotnet/machinelearning · error · ArgumentException

The data type '{dataKind}' is not handled currently.

Error message

The data type '{dataKind}' is not handled currently.

What it means

Utils.GetSymbolOfDataKind resolves a DataKind to the corresponding CSharpCodeProvider code-dom type symbol (Symbols.*). When the DataKind isn't one of the handled cases the default branch throws this ArgumentException. It's raised while GenerateClassLabels builds class/enum declarations for generated C# code, meaning the column's data type cannot be represented by the generator.

Source

Thrown at src/Microsoft.ML.CodeGenerator/Utils.cs:362

            {
                case DataKind.String:
                    return Symbols.StringSymbol;
                case DataKind.Boolean:
                    return Symbols.BoolSymbol;
                case DataKind.Single:
                    return Symbols.FloatSymbol;
                case DataKind.Double:
                    return Symbols.DoubleSymbol;
                case DataKind.Int32:
                    return Symbols.IntSymbol;
                case DataKind.UInt32:
                    return Symbols.UIntSymbol;
                case DataKind.Int64:
                    return Symbols.LongSymbol;
                case DataKind.UInt64:
                    return Symbols.UlongSymbol;
                default:
                    throw new ArgumentException($"The data type '{dataKind}' is not handled currently.");
            }
        }
    }
}

View on GitHub (pinned to 7b76e69cf9)

Solutions

1. Pre-convert the offending column to a supported numeric kind (mlContext.Transforms.Conversion.ConvertType) before code generation.
2. Upgrade Microsoft.ML.CodeGenerator to a build that maps your DataKind to a symbol.
3. Filter out or rename unsupported columns before running AutoML + codegen.
4. Extend GetSymbolOfDataKind with a case for the missing DataKind if you control the source.

Example fix

// before
var pipeline = mlContext.Transforms.Concatenate("Features", "IsWeekend", "Amount");
// after: convert bool column to a handled kind first
var pipeline = mlContext.Transforms.Conversion.ConvertType("IsWeekendNum", "IsWeekend", outputKind: DataKind.Single)
    .Append(mlContext.Transforms.Concatenate("Features", "IsWeekendNum", "Amount"));

var supported = new HashSet<DataKind> { DataKind.Single, DataKind.Double, DataKind.Int32,
    DataKind.UInt32, DataKind.Int64, DataKind.UInt64, DataKind.Int16, DataKind.UInt16, DataKind.Byte, DataKind.SByte };
foreach (var col in schema)
    if (!supported.Contains(((DataViewType)col.Type).GetRawKind().GetValueOrDefault()))
        throw new InvalidOperationException($"Column {col.Name} kind unsupported for codegen");

static bool IsSymbolSupported(DataKind kind) => kind is DataKind.Single or DataKind.Double
    or DataKind.Int32 or DataKind.UInt32 or DataKind.Int64 or DataKind.UInt64
    or DataKind.Int16 or DataKind.UInt16 or DataKind.Byte or DataKind.SByte;

try
{
    var symbol = Utils.GetSymbolOfDataKind(kind);
}
catch (ArgumentException ex) when (ex.Message.Contains("is not handled currently"))
{
    symbol = null; // skip emitting this column, log the kind
}

When it happens

Trigger: GenerateClassLabels processes a pipeline node whose column/label DataKind is outside the handled set (Boolean, DateTime, TimeSpan, String, etc.), so GetSymbolOfDataKind hits its default throw while emitting the C# class.

Common situations: Datasets with bool/datetime/string feature columns being converted to C# by mlnet codegen; version drift where newer DataKind values exist; columns loaded via custom IDataView loaders with unusual types.

Related errors

• The data type '{labelType}' is not handled currently.
• The trainer '{trainer}' is not handled currently.
• The trainer '{node.Name}' is not handled currently.
• Bad type in ColumnTypeExtensions.NumberTypeFromType: {type}
• Bad data kind in ColumnTypeExtensions.NumberTypeFromKind: {k

AI-assisted analysis of dotnet/machinelearning@7b76e69cf9 (2026-09-11). Data as JSON: /api/errors/a2b4a550827042b2. Report an issue: GitHub.