API to return ColumnPairs for a OneToOneTransformer. (#3088)

* API to return ColumnPairs for a transformer. * PR feedback. * PR feedback.
dotnet · Mar 26, 2019 · a18be69 · a18be69
1 parent 7cc1b89
commit a18be69
Show file tree

Hide file tree

Showing 4 changed files with 50 additions and 5 deletions.
diff --git a/src/Microsoft.ML.Data/Transforms/ExtensionsCatalog.cs b/src/Microsoft.ML.Data/Transforms/ExtensionsCatalog.cs
@@ -2,6 +2,7 @@
 // The .NET Foundation licenses this file to you under the MIT license.
 // See the LICENSE file in the project root for more information.
 
+using System.Collections.Generic;
 using System.Linq;
 using Microsoft.ML.Data;
 using Microsoft.ML.Runtime;
@@ -36,10 +37,12 @@ public InputOutputColumnPair(string outputColumnName, string inputColumnName = n
         }
 
         [BestFriend]
-        internal static (string outputColumnName, string inputColumnName)[] ConvertToValueTuples(InputOutputColumnPair[] infos)
-        {
-            return infos.Select(info => (info.OutputColumnName, info.InputColumnName)).ToArray();
-        }
+        internal static (string outputColumnName, string inputColumnName)[] ConvertToValueTuples(InputOutputColumnPair[] infos) =>
+            infos.Select(info => (info.OutputColumnName, info.InputColumnName)).ToArray();
+
+        [BestFriend]
+        internal static IReadOnlyList<InputOutputColumnPair> ConvertFromValueTuples((string outputColumnName, string inputColumnName)[] infos) =>
+            infos.Select(info => new InputOutputColumnPair(info.outputColumnName, info.inputColumnName)).ToList().AsReadOnly();
     }
 
     /// <summary>

diff --git a/src/Microsoft.ML.Data/Transforms/OneToOneTransformerBase.cs b/src/Microsoft.ML.Data/Transforms/OneToOneTransformerBase.cs
@@ -14,7 +14,7 @@ namespace Microsoft.ML.Data
     public abstract class OneToOneTransformerBase : RowToRowTransformerBase
     {
         [BestFriend]
-        private protected readonly (string outputColumnName, string inputColumnName)[] ColumnPairs;
+        internal readonly (string outputColumnName, string inputColumnName)[] ColumnPairs;
 
         [BestFriend]
         private protected OneToOneTransformerBase(IHost host, params (string outputColumnName, string inputColumnName)[] columns) : base(host)

diff --git a/src/Microsoft.ML.Experimental/OneToOneTransformerBaseExtensions.cs b/src/Microsoft.ML.Experimental/OneToOneTransformerBaseExtensions.cs
@@ -0,0 +1,18 @@
+// Licensed to the .NET Foundation under one or more agreements.
+// The .NET Foundation licenses this file to you under the MIT license.
+// See the LICENSE file in the project root for more information.
+
+using System.Collections.Generic;
+using Microsoft.ML.Data;
+
+namespace Microsoft.ML.Experimental
+{
+    public static class OneToOneTransformerBaseExtensions
+    {
+        /// <summary>
+        /// Returns the names of the input-output column pairs on which the transformation is applied.
+        /// </summary>
+        public static IReadOnlyList<InputOutputColumnPair> GetColumnPairs(this OneToOneTransformerBase transformer) =>
+            InputOutputColumnPair.ConvertFromValueTuples(transformer.ColumnPairs);
+    }
+}
diff --git a/test/Microsoft.ML.Tests/Transformers/NormalizerTests.cs b/test/Microsoft.ML.Tests/Transformers/NormalizerTests.cs
@@ -356,6 +356,30 @@ public void NormalizerExperimentalExtensions()
             Done();
         }
 
+        [Fact]
+        public void NormalizerExperimentalExtensionGetColumnPairs()
+        {
+            string dataPath = GetDataPath(TestDatasets.iris.trainFilename);
+
+            var loader = new TextLoader(Env, new TextLoader.Options
+            {
+                Columns = new[] {
+                    new TextLoader.Column("Label", DataKind.Single, 0),
+                    new TextLoader.Column("input", DataKind.Single, new[]{new TextLoader.Range(1, 4) }),
+                }
+            });
+
+            var data = loader.Load(dataPath);
+            var est = ML.Transforms.Normalize("output", "input", NormalizingEstimator.NormalizationMode.MinMax);
+            var t = est.Fit(data);
+
+            Assert.Single(t.GetColumnPairs());
+            Assert.Equal("output", t.GetColumnPairs()[0].OutputColumnName);
+            Assert.Equal("input", t.GetColumnPairs()[0].InputColumnName);
+
+            Done();
+        }
+
         [Fact]
         public void LpGcNormAndWhiteningWorkout()
         {