conversion of symsgdclassificationtrainer, added new estimator test, still need to pass the weight column

Author: artidoro

src/Microsoft.ML.HalLearners/SymSgdClassificationTrainer.cs

@@ -6,6 +6,7 @@
 using System.Collections.Generic;
 using System.Runtime.InteropServices;
 using System.Security;
+using Microsoft.ML.Core.Data;
 using Microsoft.ML.Runtime;
 using Microsoft.ML.Runtime.CommandLine;
 using Microsoft.ML.Runtime.Data;
@@ -131,7 +132,7 @@ private RoleMappedData PrepareDataFromTrainingExamples(IChannel ch, RoleMappedDa
             return examplesToFeedTrain;
         }
 
-        public override TPredictor Train(TrainContext context)
+        protected override TPredictor TrainModelCore(TrainContext context)
         {
             Host.CheckValue(context, nameof(context));
             TPredictor pred;
@@ -161,29 +162,33 @@ public override TPredictor Train(TrainContext context)
         /// <param name="advancedSettings">A delegate to apply all the advanced arguments to the algorithm.</param>
         public SymSgdClassificationTrainer(IHostEnvironment env, string featureColumn, string labelColumn,
             string weightColumn = null, Action<Arguments> advancedSettings = null)
-            : this(env, ArgsInit(featureColumn, labelColumn, weightColumn, advancedSettings))
+            : base(Contracts.CheckRef(env, nameof(env)).Register(LoadNameValue), TrainerUtils.MakeR4VecFeature(featureColumn),
+                  TrainerUtils.MakeR4ScalarLabel(labelColumn), TrainerUtils.MakeR4ScalarWeightColumn(weightColumn))
         {
-        }
+            var args = new Arguments();
 
-        public SymSgdClassificationTrainer(IHostEnvironment env, Arguments args)
-            : base(env, LoadNameValue)
-        {
+            //apply the advanced args, if the user supplied any
             args.Check(Host);
+            advancedSettings?.Invoke(args);
+            args.FeatureColumn = featureColumn;
+            args.LabelColumn = labelColumn;
+            // TODO:
+            //args.WeightColumn = weightColumn;
+
             _args = args;
             Info = new TrainerInfo();
         }
 
-        private static Arguments ArgsInit(string featureColumn, string labelColumn,
-            string weightColumn, Action<Arguments> advancedSettings)
+        /// <summary>
+        /// Initializes a new instance of <see cref="SymSgdClassificationTrainer"/>
+        /// </summary>
+        internal SymSgdClassificationTrainer(IHostEnvironment env, Arguments args)
+            : base(Contracts.CheckRef(env, nameof(env)).Register(LoadNameValue), TrainerUtils.MakeR4VecFeature(args.FeatureColumn),
+                  TrainerUtils.MakeR4ScalarLabel(args.LabelColumn), TrainerUtils.MakeR4ScalarWeightColumn(null))
         {
-            var args = new Arguments();
-
-            //apply the advanced args, if the user supplied any
-            advancedSettings?.Invoke(args);
-            args.FeatureColumn = featureColumn;
-            args.LabelColumn = labelColumn;
-            args.WeightColumn = weightColumn;
-            return args;
+            args.Check(Host);
+            _args = args;
+            Info = new TrainerInfo();
         }
 
         private TPredictor CreatePredictor(VBuffer<float> weights, float bias)
@@ -197,6 +202,19 @@ private TPredictor CreatePredictor(VBuffer<float> weights, float bias)
             return new ParameterMixingCalibratedPredictor(Host, predictor, new PlattCalibrator(Host, -1, 0));
         }
 
+        protected override BinaryPredictionTransformer<TPredictor> MakeTransformer(TPredictor model, ISchema trainSchema)
+             => new BinaryPredictionTransformer<TPredictor>(Host, model, trainSchema, FeatureColumn.Name);
+
+        protected override SchemaShape.Column[] GetOutputColumnsCore(SchemaShape inputSchema)
+        {
+            return new[]
+            {
+                new SchemaShape.Column(DefaultColumnNames.Score, SchemaShape.Column.VectorKind.Scalar, NumberType.R4, false, new SchemaShape(MetadataUtils.GetTrainerOutputMetadata())),
+                new SchemaShape.Column(DefaultColumnNames.Probability, SchemaShape.Column.VectorKind.Scalar, NumberType.R4, false, new SchemaShape(MetadataUtils.GetTrainerOutputMetadata(true))),
+                new SchemaShape.Column(DefaultColumnNames.PredictedLabel, SchemaShape.Column.VectorKind.Scalar, BoolType.Instance, false, new SchemaShape(MetadataUtils.GetTrainerOutputMetadata()))
+            };
+        }
+
         [TlcModule.EntryPoint(Name = "Trainers.SymSgdBinaryClassifier",
             Desc = "Train a symbolic SGD.",
             UserName = SymSgdClassificationTrainer.UserNameValue,

test/Microsoft.ML.Tests/TrainerEstimators/SymSgdClassificationTests.cs

@@ -0,0 +1,37 @@
+// 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 Microsoft.ML.Runtime.Data;
+using Microsoft.ML.Runtime.RunTests;
+using Microsoft.ML.Runtime.SymSgd;
+using Xunit;
+
+namespace Microsoft.ML.Tests.TrainerEstimators
+{
+    public partial class TrainerEstimators
+    {
+        private IDataView GetBreastCancerDataview()
+        {
+            return new TextLoader(Env,
+            new TextLoader.Arguments()
+            {
+                HasHeader = true,
+                Column = new[]
+                {
+                    new TextLoader.Column("Label", DataKind.R4, 0),
+                    new TextLoader.Column("Features", DataKind.R4, new [] { new TextLoader.Range(1, 9) } )
+                }
+            }).Read(new MultiFileSource(GetDataPath(TestDatasets.breastCancer.trainFilename)));
+        }
+
+        [Fact]
+        public void TestEstimatorSymSgdClassificationTrainer()
+        {
+            var dataView = GetBreastCancerDataview();
+            var pipe = new SymSgdClassificationTrainer(Env, "Features", "Label");
+            TestEstimatorCore(pipe, dataView);
+            Done();
+        }
+    }
+}