dotnet/machinelearning · error · ArgumentException

Provided {columnPurpose} column '{columnName}' was of type {

Error message

Provided {columnPurpose} column '{columnName}' was of type {itemType}, but only types {string.Join(", ", allowedTypes)} are allowed.

What it means

The multi-type branch of ValidateTrainDataColumn's type check: when the column's item type is not among the several allowedTypes, this ArgumentException lists the actual type and all allowed types. Same root cause as the single-type variant, just with a richer allowed set.

Source

Thrown at src/Microsoft.ML.AutoML/Utils/UserInputValidationUtil.cs:278

                throw new ArgumentException(exceptionMessage);
            }

            if (allowedTypes == null)
            {
                return;
            }
            var column = nullableColumn.Value;
            var itemType = column.Type.GetItemType();
            if (!allowedTypes.Contains(itemType))
            {
                if (allowedTypes.Count() == 1)
                {
                    throw new ArgumentException($"Provided {columnPurpose} column '{columnName}' was of type {itemType}, " +
                        $"but only type {allowedTypes.First()} is allowed.");
                }
                else
                {
                    throw new ArgumentException($"Provided {columnPurpose} column '{columnName}' was of type {itemType}, " +
                        $"but only types {string.Join(", ", allowedTypes)} are allowed.");
                }
            }
        }

        private static string ClosestNamed(IDataView trainData, string columnName, int maxAllowableEditDistance = int.MaxValue)
        {
            var minEditDistance = int.MaxValue;
            var closestNamed = string.Empty;
            foreach (var column in trainData.Schema)
            {
                var editDistance = StringEditDistance.GetLevenshteinDistance(column.Name, columnName);
                if (editDistance < minEditDistance)
                {
                    minEditDistance = editDistance;
                    closestNamed = column.Name;
                }
            }

View on GitHub (pinned to 7b76e69cf9)

Solutions

  1. Convert the column to one of the listed allowed types (ConvertType transform) before Execute.
  2. Correct the ColumnInformation entry to a column whose type matches the allowed list.
  3. Use AutoML's InferColumns/InferColumnInformation so column purposes and types are inferred consistently.

Example fix

// before
var colInfo = new ColumnInformation { LabelColumnName = "Date" }; // DateTime, task allows Single
// after
var typed = mlContext.Transforms.Conversion.ConvertType("Date", outputKind: DataKind.Single).Fit(trainData).Transform(trainData);
Defensive patterns

Strategy: validation

Validate before calling

var itemType = trainData.Schema[labelCol].Type.GetItemType();
var allowed = new[] { NumberDataViewType.Single, /* task-specific types */ };
if (!allowed.Contains(itemType)) throw new InvalidOperationException($"{itemType} not allowed for label");

Try / catch

try { result = experiment.Execute(trainData, settings, columnInfo); }
catch (ArgumentException ex) when (ex.Message.Contains("only types")) { /* convert to one of the listed types */ }

Prevention

When it happens

Trigger: Designating a label column whose item type isn't in the task's allowed set (e.g. classification allows Single and NumberDataViewType key types — a Double or DateTime label fails).

Common situations: DateTime or Double labels for a task expecting Single/keys; label loaded as string for multiclass that requires KeyType; wrong column designated as label.

Understand the failure class

Background: Type mismatch errors: IllegalArgumentException, TypeError and type guards across 150 open-source libraries — this error's family across 150 libraries.

Related errors

AI-assisted analysis of dotnet/machinelearning@7b76e69cf9 (2026-09-11). Data as JSON: /api/errors/e4a0edf61c23081d. Report an issue: GitHub.