Improve exception msg by adding column name

src/Microsoft.ML.Data/Transforms/KeyToVector.cs

@@ -246,14 +246,28 @@ public Mapper(KeyToVectorMappingTransformer parent, DataViewSchema inputSchema)
                 _parent = parent;
                 _infos = CreateInfos(inputSchema);
                 _types = new VectorDataViewType[_parent.ColumnPairs.Length];
-                for (int i = 0; i < _parent.ColumnPairs.Length; i++)
+
+                // The following try catch block is designed to provide user a better exception message
+                // by providing related column name when some exceptions occur in VectorDataViewType()(e.g overflow)
+                // This change is related with https://github.com/dotnet/machinelearning/issues/5211
+                try
                 {
-                    int valueCount = _infos[i].TypeSrc.GetValueCount();
-                    int keyCount = _infos[i].TypeSrc.GetItemType().GetKeyCountAsInt32(Host);
-                    if (_parent._columns[i].OutputCountVector || valueCount == 1)
-                        _types[i] = new VectorDataViewType(NumberDataViewType.Single, keyCount);
-                    else
-                        _types[i] = new VectorDataViewType(NumberDataViewType.Single, valueCount, keyCount);
+                    for (int i = 0; i < _parent.ColumnPairs.Length; i++)
+                    {
+                        int valueCount = _infos[i].TypeSrc.GetValueCount();
+                        int keyCount = _infos[i].TypeSrc.GetItemType().GetKeyCountAsInt32(Host);
+                        if (_parent._columns[i].OutputCountVector || valueCount == 1)
+                            _types[i] = new VectorDataViewType(NumberDataViewType.Single, keyCount);
+                        else
+                            _types[i] = new VectorDataViewType(NumberDataViewType.Single, valueCount, keyCount);
+                    }
+                }
+                catch (Exception e)
+                {
+                    var errorMsg = e.Message + " Related column: ";
+                    foreach (var info in _infos)
+                        errorMsg += info.Name + " ";
+                    throw Host.Except(errorMsg);
                 }
             }
 

test/Microsoft.ML.Tests/Transformers/KeyToVectorEstimatorTests.cs

@@ -3,6 +3,7 @@
 // See the LICENSE file in the project root for more information.
 
 using System;
+using System.Collections.Generic;
 using System.IO;
 using System.Linq;
 using Microsoft.ML.Data;
@@ -222,5 +223,51 @@ public void TestOldSavingAndLoading()
                 var loadedView = ModelFileUtils.LoadTransforms(Env, dataView, ms);
             }
         }
+
+        public class ModelInput
+        {
+            [ColumnName("Label"), LoadColumn(0)]
+            public int Label { get; set; }
+
+
+            [ColumnName("ProblematicColumn"), LoadColumn(1)]
+            public string ProblematicColumn { get; set; }
+        }
+
+        static IEnumerable<ModelInput> GetData()
+        {
+            for (int i = 0; i < 1000; i++)
+            {
+                yield return new ModelInput { Label = i % 3, ProblematicColumn = (i % 200).ToString() };
+            }
+        }
+
+        [Fact]
+        public void KeyToVectorOverflowTest()
+        {
+            // This test is introduced for https://github.com/dotnet/machinelearning/issues/5211
+            // that provides users an informational exception message
+            MLContext mlContext = new MLContext(1);
+
+            IDataView dataview = mlContext.Data.LoadFromEnumerable(GetData());
+
+            var pipeline = mlContext.Transforms.Conversion.MapValueToKey("Label")
+                   .Append(mlContext.Transforms.Categorical.OneHotHashEncoding("ProblematicColumn"))
+                   .Append(mlContext.Transforms.CopyColumns("Features", "ProblematicColumn"));
+
+            var featurizedData = pipeline.Fit(dataview).Transform(dataview);
+
+            try
+            {
+                var transformer = pipeline.Fit(featurizedData);
+                Assert.True(false);
+            }
+            catch (System.Exception ex)
+            {
+                Assert.Contains("Arithmetic operation resulted in an overflow. Related column: ProblematicColumn", ex.Message);
+                return;
+            }
+
+        }
     }
 }