Added new test to columninfo and fixing up API (dotnet#178)

Author: vinodshanbhag

src/Microsoft.ML.Auto/API/BinaryClassificationExperiment.cs

@@ -49,7 +49,7 @@ public IEnumerable<RunResult<BinaryClassificationMetrics>> Execute(IDataView tra
             return Execute(_context, trainData, columnInformation, validationData, preFeaturizers);
         }
 
-        internal RunResult<BinaryClassificationMetrics> Execute(IDataView trainData, uint numberOfCVFolds, ColumnInformation columnInformation = null, IEstimator<ITransformer> preFeaturizers = null)
+        internal IEnumerable<RunResult<BinaryClassificationMetrics>> Execute(IDataView trainData, uint numberOfCVFolds, ColumnInformation columnInformation = null, IEstimator<ITransformer> preFeaturizers = null)
         {
             throw new NotImplementedException();
         }

src/Microsoft.ML.Auto/API/MulticlassClassificationExperiment.cs

@@ -49,7 +49,7 @@ public IEnumerable<RunResult<MultiClassClassifierMetrics>> Execute(IDataView tra
             return Execute(_context, trainData, columnInformation, validationData, preFeaturizers);
         }
 
-        internal RunResult<BinaryClassificationMetrics> Execute(IDataView trainData, uint numberOfCVFolds, ColumnInformation columnInformation = null, IEstimator<ITransformer> preFeaturizers = null)
+        internal IEnumerable<RunResult<MultiClassClassifierMetrics>> Execute(IDataView trainData, uint numberOfCVFolds, ColumnInformation columnInformation = null, IEstimator<ITransformer> preFeaturizers = null)
         {
             throw new NotImplementedException();
         }

src/Microsoft.ML.Auto/API/RegressionExperiment.cs

@@ -52,7 +52,7 @@ public IEnumerable<RunResult<RegressionMetrics>> Execute(IDataView trainData, ID
             return Execute(_context, trainData, columnInformation, validationData, preFeaturizers);
         }
 
-        internal RunResult<RegressionMetrics> Execute(IDataView trainData, uint numberOfCVFolds, ColumnInformation columnInformation = null, IEstimator<ITransformer> preFeaturizers = null)
+        internal IEnumerable<RunResult<RegressionMetrics>> Execute(IDataView trainData, uint numberOfCVFolds, ColumnInformation columnInformation = null, IEstimator<ITransformer> preFeaturizers = null)
         {
             throw new NotImplementedException();
         }

src/Test/ColumnInferenceTests.cs

@@ -9,7 +9,7 @@ namespace Microsoft.ML.Auto.Test
     public class ColumnInferenceTests
     {
         [TestMethod]
-        public void UnGroupColumnsTest()
+        public void UnGroupReturnsMoreColumnsThanGroup()
         {
             var dataPath = DatasetUtil.DownloadUciAdultDataset();
             var context = new MLContext();
@@ -24,7 +24,7 @@ public void UnGroupColumnsTest()
         }
 
         [TestMethod]
-        public void IncorrectLabelColumnTest()
+        public void IncorrectLabelColumnThrows()
         {
             var dataPath = DatasetUtil.DownloadUciAdultDataset();
             var context = new MLContext();
@@ -33,13 +33,13 @@ public void IncorrectLabelColumnTest()
 
         [TestMethod]
         [ExpectedException(typeof(ArgumentOutOfRangeException))]
-        public void InferColumnsLabelIndexOutOfBounds()
+        public void LabelIndexOutOfBoundsThrows()
         {
             new MLContext().AutoInference().InferColumns(DatasetUtil.DownloadUciAdultDataset(), 100);
         }
 
         [TestMethod]
-        public void InferColumnsLabelIndex()
+        public void IdentifyLabelColumnThroughIndexWithHeader()
         {
             var result = new MLContext().AutoInference().InferColumns(DatasetUtil.DownloadUciAdultDataset(), 14, hasHeader: true);
             Assert.AreEqual(true, result.TextLoaderArgs.HasHeader);
@@ -49,7 +49,7 @@ public void InferColumnsLabelIndex()
         }
 
         [TestMethod]
-        public void InferColumnsLabelIndexNoHeaders()
+        public void IdentifyLabelColumnThroughIndexWithoutHeader()
         {
             var result = new MLContext().AutoInference().InferColumns(DatasetUtil.DownloadIrisDataset(), DatasetUtil.IrisDatasetLabelColIndex);
             Assert.AreEqual(false, result.TextLoaderArgs.HasHeader);
@@ -60,15 +60,15 @@ public void InferColumnsLabelIndexNoHeaders()
         }
 
         [TestMethod]
-        public void InferColumnsWithDatasetWithEmptyColumn()
+        public void DatasetWithEmptyColumn()
         {
             var result = new MLContext().AutoInference().InferColumns(@".\TestData\DatasetWithEmptyColumn.txt", DefaultColumnNames.Label);
             var emptyColumn = result.TextLoaderArgs.Column.First(c => c.Name == "Empty");
             Assert.AreEqual(DataKind.TX, emptyColumn.Type);
         }
 
         [TestMethod]
-        public void InferColumnsWithDatasetWithBoolColumn()
+        public void DatasetWithBoolColumn()
         {
             var result = new MLContext().AutoInference().InferColumns(@".\TestData\BinaryDatasetWithBoolColumn.txt", DefaultColumnNames.Label);
             Assert.AreEqual(2, result.TextLoaderArgs.Column.Count());
@@ -86,7 +86,7 @@ public void InferColumnsWithDatasetWithBoolColumn()
         }
 
         [TestMethod]
-        public void InferColumnsWhereNameColumnIsOnlyFeature()
+        public void WhereNameColumnIsOnlyFeature()
         {
             var result = new MLContext().AutoInference().InferColumns(@".\TestData\NameColumnIsOnlyFeatureDataset.txt", DefaultColumnNames.Label);
             Assert.AreEqual(2, result.TextLoaderArgs.Column.Count());
@@ -100,5 +100,17 @@ public void InferColumnsWhereNameColumnIsOnlyFeature()
             Assert.AreEqual("Username", result.ColumnInformation.TextColumns.First());
             Assert.AreEqual(DefaultColumnNames.Label, result.ColumnInformation.LabelColumn);
         }
+
+        [TestMethod]
+        public void DefaultColumnNamesInferredCorrectly()
+        {
+            var result = new MLContext().AutoInference().InferColumns(@".\TestData\DatasetWithDefaultColumnNames.txt", DefaultColumnNames.Label, groupColumns : false);
+
+            Assert.AreEqual(DefaultColumnNames.Label, result.ColumnInformation.LabelColumn);
+            Assert.AreEqual(DefaultColumnNames.Name, result.ColumnInformation.NameColumn);
+            Assert.AreEqual(DefaultColumnNames.Weight, result.ColumnInformation.WeightColumn);
+            Assert.AreEqual(DefaultColumnNames.GroupId, result.ColumnInformation.GroupIdColumn);
+            Assert.AreEqual(result.ColumnInformation.NumericColumns.Count(), 3);
+        }
     }
 }

src/Test/Test.csproj

@@ -19,6 +19,9 @@
   </ItemGroup>
 
   <ItemGroup>
+    <None Update="TestData\DatasetWithDefaultColumnNames.txt">
+      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+    </None>
     <None Update="TestData\NameColumnIsOnlyFeatureDataset.txt">
       <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
     </None>

src/Test/TestData/DatasetWithDefaultColumnNames.txt

@@ -0,0 +1,4 @@
+Label,GroupId,Weight,Name,Features,FeatureContributions,Feature1
+0,2,1,GUID1,1,1,1
+0,4,1,GUID2,1,1,1
+1,1,1,GUID3,1,1,1