Trainer estimator cleanup for FastTrees and LightGBM

src/Microsoft.ML.Data/Training/TrainerEstimatorBase.cs

@@ -41,6 +41,11 @@ public abstract class TrainerEstimatorBase<TTransformer, TModel> : ITrainerEstim
         /// </summary>
         public readonly SchemaShape.Column WeightColumn;
 
+        /// <summary>
+        /// The optional groupID column that the ranking trainers expects.
+        /// </summary>
+        public readonly SchemaShape.Column GroupIdColumn;
+
         protected readonly IHost Host;
 
         /// <summary>
@@ -50,17 +55,23 @@ public abstract class TrainerEstimatorBase<TTransformer, TModel> : ITrainerEstim
 
         public abstract PredictionKind PredictionKind { get; }
 
-        public TrainerEstimatorBase(IHost host, SchemaShape.Column feature, SchemaShape.Column label, SchemaShape.Column weight = null)
+        public TrainerEstimatorBase(IHost host,
+            SchemaShape.Column feature,
+            SchemaShape.Column label,
+            SchemaShape.Column weight = null,
+            SchemaShape.Column groupId = null)
         {
             Contracts.CheckValue(host, nameof(host));
             Host = host;
             Host.CheckValue(feature, nameof(feature));
             Host.CheckValueOrNull(label);
             Host.CheckValueOrNull(weight);
+            Host.CheckValueOrNull(groupId);
 
             FeatureColumn = feature;
             LabelColumn = label;
             WeightColumn = weight;
+            GroupIdColumn = groupId;
         }
 
         public TTransformer Fit(IDataView input) => TrainTransformer(input);
@@ -150,7 +161,7 @@ protected TTransformer TrainTransformer(IDataView trainSet,
         protected abstract TTransformer MakeTransformer(TModel model, Schema trainSchema);
 
         private RoleMappedData MakeRoles(IDataView data) =>
-            new RoleMappedData(data, label: LabelColumn?.Name, feature: FeatureColumn.Name, weight: WeightColumn?.Name);
+            new RoleMappedData(data, label: LabelColumn?.Name, feature: FeatureColumn.Name, group: GroupIdColumn?.Name, weight: WeightColumn?.Name);
 
         IPredictor ITrainer.Train(TrainContext context) => Train(context);
     }

src/Microsoft.ML.Data/Training/TrainerUtils.cs

@@ -362,9 +362,14 @@ public static SchemaShape.Column MakeR4ScalarLabel(string labelColumn)
         /// <summary>
         /// The <see cref="SchemaShape.Column"/> for the label column for regression tasks.
         /// </summary>
-        /// <param name="labelColumn">name of the weight column</param>
-        public static SchemaShape.Column MakeU4ScalarLabel(string labelColumn)
-            => new SchemaShape.Column(labelColumn, SchemaShape.Column.VectorKind.Scalar, NumberType.U4, true);
+        /// <param name="columnName">name of the weight column</param>
+        public static SchemaShape.Column MakeU4ScalarColumn(string columnName)
+        {
+            if (columnName == null)
+                return null;
+
+            return new SchemaShape.Column(columnName, SchemaShape.Column.VectorKind.Scalar, NumberType.U4, true);
+        }
 
         /// <summary>
         /// The <see cref="SchemaShape.Column"/> for the feature column.
@@ -377,9 +382,10 @@ public static SchemaShape.Column MakeR4VecFeature(string featureColumn)
         /// The <see cref="SchemaShape.Column"/> for the weight column.
         /// </summary>
         /// <param name="weightColumn">name of the weight column</param>
-        public static SchemaShape.Column MakeR4ScalarWeightColumn(string weightColumn)
+        /// <param name="isExplicit">whether the column is implicitly, or explicitly defined</param>
+        public static SchemaShape.Column MakeR4ScalarWeightColumn(string weightColumn, bool isExplicit = true)
         {
-            if (weightColumn == null)
+            if (weightColumn == null || !isExplicit)
                 return null;
             return new SchemaShape.Column(weightColumn, SchemaShape.Column.VectorKind.Scalar, NumberType.R4, false);
         }

src/Microsoft.ML.FastTree/FastTree.cs

@@ -94,7 +94,7 @@ public abstract class FastTreeTrainerBase<TArgs, TTransformer, TModel> :
         /// </summary>
         private protected FastTreeTrainerBase(IHostEnvironment env, SchemaShape.Column label, string featureColumn,
             string weightColumn = null, string groupIdColumn = null, Action<TArgs> advancedSettings = null)
-            : base(Contracts.CheckRef(env, nameof(env)).Register(RegisterName), TrainerUtils.MakeR4VecFeature(featureColumn), label, TrainerUtils.MakeR4ScalarWeightColumn(weightColumn))
+            : base(Contracts.CheckRef(env, nameof(env)).Register(RegisterName), TrainerUtils.MakeR4VecFeature(featureColumn), label, TrainerUtils.MakeR4ScalarWeightColumn(weightColumn), TrainerUtils.MakeU4ScalarColumn(groupIdColumn))
         {
             Args = new TArgs();
 
@@ -128,7 +128,7 @@ private protected FastTreeTrainerBase(IHostEnvironment env, SchemaShape.Column l
         /// Legacy constructor that is used when invoking the classes deriving from this, through maml.
         /// </summary>
         private protected FastTreeTrainerBase(IHostEnvironment env, TArgs args, SchemaShape.Column label)
-            : base(Contracts.CheckRef(env, nameof(env)).Register(RegisterName), TrainerUtils.MakeR4VecFeature(args.FeatureColumn), label, TrainerUtils.MakeR4ScalarWeightColumn(args.WeightColumn))
+            : base(Contracts.CheckRef(env, nameof(env)).Register(RegisterName), TrainerUtils.MakeR4VecFeature(args.FeatureColumn), label, TrainerUtils.MakeR4ScalarWeightColumn(args.WeightColumn, args.WeightColumn.IsExplicit))
         {
             Host.CheckValue(args, nameof(args));
             Args = args;

src/Microsoft.ML.FastTree/FastTreeTweedie.cs

@@ -35,10 +35,10 @@ namespace Microsoft.ML.Trainers.FastTree
     public sealed partial class FastTreeTweedieTrainer
          : BoostingFastTreeTrainerBase<FastTreeTweedieTrainer.Arguments, RegressionPredictionTransformer<FastTreeTweediePredictor>, FastTreeTweediePredictor>
     {
-        public const string LoadNameValue = "FastTreeTweedieRegression";
-        public const string UserNameValue = "FastTree (Boosted Trees) Tweedie Regression";
-        public const string Summary = "Trains gradient boosted decision trees to fit target values using a Tweedie loss function. This learner is a generalization of Poisson, compound Poisson, and gamma regression.";
-        public const string ShortName = "fttweedie";
+        internal const string LoadNameValue = "FastTreeTweedieRegression";
+        internal const string UserNameValue = "FastTree (Boosted Trees) Tweedie Regression";
+        internal const string Summary = "Trains gradient boosted decision trees to fit target values using a Tweedie loss function. This learner is a generalization of Poisson, compound Poisson, and gamma regression.";
+        internal const string ShortName = "fttweedie";
 
         private TestHistory _firstTestSetHistory;
         private Test _trainRegressionTest;

src/Microsoft.ML.FastTree/GamTrainer.cs

@@ -154,7 +154,7 @@ private protected GamTrainerBase(IHostEnvironment env, string name, SchemaShape.
 
         private protected GamTrainerBase(IHostEnvironment env, TArgs args, string name, SchemaShape.Column label)
             : base(Contracts.CheckRef(env, nameof(env)).Register(name), TrainerUtils.MakeR4VecFeature(args.FeatureColumn),
-                  label, TrainerUtils.MakeR4ScalarWeightColumn(args.WeightColumn))
+                  label, TrainerUtils.MakeR4ScalarWeightColumn(args.WeightColumn, args.WeightColumn.IsExplicit))
         {
             Contracts.CheckValue(env, nameof(env));
             Host.CheckValue(args, nameof(args));

src/Microsoft.ML.FastTree/FastTreeCatalog.cs → src/Microsoft.ML.FastTree/TreeTrainersCatalog.cs

@@ -12,7 +12,7 @@ namespace Microsoft.ML
     /// <summary>
     /// FastTree <see cref="TrainContextBase"/> extension methods.
     /// </summary>
-    public static class FastTreeRegressionExtensions
+    public static class TreeExtensions
     {
         /// <summary>
         /// Predict a target using a decision tree regression model trained with the <see cref="FastTreeRegressionTrainer"/>.
@@ -40,10 +40,6 @@ public static FastTreeRegressionTrainer FastTree(this RegressionContext.Regressi
             var env = CatalogUtils.GetEnvironment(ctx);
             return new FastTreeRegressionTrainer(env, label, features, weights, numLeaves, numTrees, minDatapointsInLeafs, learningRate, advancedSettings);
         }
-    }
-
-    public static class FastTreeBinaryClassificationExtensions
-    {
 
         /// <summary>
         /// Predict a target using a decision tree binary classification model trained with the <see cref="FastTreeBinaryClassificationTrainer"/>.
@@ -71,10 +67,6 @@ public static FastTreeBinaryClassificationTrainer FastTree(this BinaryClassifica
             var env = CatalogUtils.GetEnvironment(ctx);
             return new FastTreeBinaryClassificationTrainer(env, label, features, weights, numLeaves, numTrees, minDatapointsInLeafs, learningRate, advancedSettings);
         }
-    }
-
-    public static class FastTreeRankingExtensions
-    {
 
         /// <summary>
         /// Ranks a series of inputs based on their relevance, training a decision tree ranking model through the <see cref="FastTreeRankingTrainer"/>.
@@ -96,5 +88,62 @@ public static FastTreeRankingTrainer FastTree(this RankingContext.RankingTrainer
             var env = CatalogUtils.GetEnvironment(ctx);
             return new FastTreeRankingTrainer(env, label, features, groupId, weights, advancedSettings);
         }
+
+        /// <summary>
+        /// Predict a target using a decision tree regression model trained with the <see cref="FastTreeRegressionTrainer"/>.
+        /// </summary>
+        /// <param name="ctx">The <see cref="RegressionContext"/>.</param>
+        /// <param name="label">The label column.</param>
+        /// <param name="features">The features colum.</param>
+        /// <param name="weights">The optional weights column.</param>
+        /// <param name="advancedSettings">Algorithm advanced settings.</param>
+        public static BinaryClassificationGamTrainer GeneralizedAdditiveMethods(this RegressionContext.RegressionTrainers ctx,
+            string label = DefaultColumnNames.Label,
+            string features = DefaultColumnNames.Features,
+            string weights = null,
+            Action<BinaryClassificationGamTrainer.Arguments> advancedSettings = null)
+        {
+            Contracts.CheckValue(ctx, nameof(ctx));
+            var env = CatalogUtils.GetEnvironment(ctx);
+            return new BinaryClassificationGamTrainer(env, label, features, weights, advancedSettings);
+        }
+
+        /// <summary>
+        /// Predict a target using a decision tree binary classification model trained with the <see cref="FastTreeBinaryClassificationTrainer"/>.
+        /// </summary>
+        /// <param name="ctx">The <see cref="BinaryClassificationContext"/>.</param>
+        /// <param name="label">The label column.</param>
+        /// <param name="features">The features colum.</param>
+        /// <param name="weights">The optional weights column.</param>
+        /// <param name="advancedSettings">Algorithm advanced settings.</param>
+        public static RegressionGamTrainer GeneralizedAdditiveMethods(this BinaryClassificationContext.BinaryClassificationTrainers ctx,
+            string label = DefaultColumnNames.Label,
+            string features = DefaultColumnNames.Features,
+            string weights = null,
+            Action<RegressionGamTrainer.Arguments> advancedSettings = null)
+        {
+            Contracts.CheckValue(ctx, nameof(ctx));
+            var env = CatalogUtils.GetEnvironment(ctx);
+            return new RegressionGamTrainer(env, label, features, weights, advancedSettings);
+        }
+
+        /// <summary>
+        /// Predict a target using a decision tree regression model trained with the <see cref="FastTreeTweedieTrainer"/>.
+        /// </summary>
+        /// <param name="ctx">The <see cref="RegressionContext"/>.</param>
+        /// <param name="label">The label column.</param>
+        /// <param name="features">The features colum.</param>
+        /// <param name="weights">The optional weights column.</param>
+        /// <param name="advancedSettings">Algorithm advanced settings.</param>
+        public static FastTreeTweedieTrainer FastTreeTweedie(this RegressionContext.RegressionTrainers ctx,
+            string label = DefaultColumnNames.Label,
+            string features = DefaultColumnNames.Features,
+            string weights = null,
+            Action<FastTreeTweedieTrainer.Arguments> advancedSettings = null)
+        {
+            Contracts.CheckValue(ctx, nameof(ctx));
+            var env = CatalogUtils.GetEnvironment(ctx);
+            return new FastTreeTweedieTrainer(env, label, features, weights, advancedSettings: advancedSettings);
+        }
     }
 }

src/Microsoft.ML.FastTree/FastTreeStatic.cs → src/Microsoft.ML.FastTree/TreeTrainersStatic.cs

@@ -14,7 +14,7 @@ namespace Microsoft.ML.StaticPipe
     /// <summary>
     /// FastTree <see cref="TrainContextBase"/> extension methods.
     /// </summary>
-    public static class FastTreeRegressionExtensions
+    public static class TreeRegressionExtensions
     {
         /// <summary>
         /// FastTree <see cref="RegressionContext"/> extension method.
@@ -50,7 +50,7 @@ public static Scalar<float> FastTree(this RegressionContext.RegressionTrainers c
             Action<FastTreeRegressionTrainer.Arguments> advancedSettings = null,
             Action<FastTreeRegressionPredictor> onFit = null)
         {
-            FastTreeStaticsUtils.CheckUserValues(label, features, weights, numLeaves, numTrees, minDatapointsInLeafs, learningRate, advancedSettings, onFit);
+            CheckUserValues(label, features, weights, numLeaves, numTrees, minDatapointsInLeafs, learningRate, advancedSettings, onFit);
 
             var rec = new TrainerEstimatorReconciler.Regression(
                (env, labelName, featuresName, weightsName) =>
@@ -64,10 +64,6 @@ public static Scalar<float> FastTree(this RegressionContext.RegressionTrainers c
 
             return rec.Score;
         }
-    }
-
-    public static class FastTreeBinaryClassificationExtensions
-    {
 
         /// <summary>
         /// FastTree <see cref="BinaryClassificationContext"/> extension method.
@@ -98,7 +94,7 @@ public static (Scalar<float> score, Scalar<float> probability, Scalar<bool> pred
             Action<FastTreeBinaryClassificationTrainer.Arguments> advancedSettings = null,
             Action<IPredictorWithFeatureWeights<float>> onFit = null)
         {
-            FastTreeStaticsUtils.CheckUserValues(label, features, weights, numLeaves, numTrees, minDatapointsInLeafs, learningRate, advancedSettings, onFit);
+            CheckUserValues(label, features, weights, numLeaves, numTrees, minDatapointsInLeafs, learningRate, advancedSettings, onFit);
 
             var rec = new TrainerEstimatorReconciler.BinaryClassifier(
                (env, labelName, featuresName, weightsName) =>
@@ -114,10 +110,6 @@ public static (Scalar<float> score, Scalar<float> probability, Scalar<bool> pred
 
             return rec.Output;
         }
-    }
-
-    public static class FastTreeRankingExtensions
-    {
 
         /// <summary>
         /// FastTree <see cref="RankingContext"/>.
@@ -148,7 +140,7 @@ public static Scalar<float> FastTree<TVal>(this RankingContext.RankingTrainers c
             Action<FastTreeRankingTrainer.Arguments> advancedSettings = null,
             Action<FastTreeRankingPredictor> onFit = null)
         {
-            FastTreeStaticsUtils.CheckUserValues(label, features, weights, numLeaves, numTrees, minDatapointsInLeafs, learningRate, advancedSettings, onFit);
+            CheckUserValues(label, features, weights, numLeaves, numTrees, minDatapointsInLeafs, learningRate, advancedSettings, onFit);
 
             var rec = new TrainerEstimatorReconciler.Ranker<TVal>(
                (env, labelName, featuresName, groupIdName, weightsName) =>
@@ -161,17 +153,14 @@ public static Scalar<float> FastTree<TVal>(this RankingContext.RankingTrainers c
 
             return rec.Score;
         }
-    }
 
-    internal class FastTreeStaticsUtils
-    {
         internal static void CheckUserValues(PipelineColumn label, Vector<float> features, Scalar<float> weights,
-        int numLeaves,
-        int numTrees,
-        int minDatapointsInLeafs,
-        double learningRate,
-        Delegate advancedSettings,
-        Delegate onFit)
+            int numLeaves,
+            int numTrees,
+            int minDatapointsInLeafs,
+            double learningRate,
+            Delegate advancedSettings,
+            Delegate onFit)
         {
             Contracts.CheckValue(label, nameof(label));
             Contracts.CheckValue(features, nameof(features));

src/Microsoft.ML.HalLearners/OlsLinearRegression.cs

@@ -85,7 +85,7 @@ public OlsLinearRegressionTrainer(IHostEnvironment env, string featureColumn, st
         /// </summary>
         internal OlsLinearRegressionTrainer(IHostEnvironment env, Arguments args)
             : base(Contracts.CheckRef(env, nameof(env)).Register(LoadNameValue), TrainerUtils.MakeR4VecFeature(args.FeatureColumn),
-                  TrainerUtils.MakeR4ScalarLabel(args.LabelColumn), TrainerUtils.MakeR4ScalarWeightColumn(args.WeightColumn))
+                  TrainerUtils.MakeR4ScalarLabel(args.LabelColumn), TrainerUtils.MakeR4ScalarWeightColumn(args.WeightColumn, args.WeightColumn.IsExplicit))
         {
             Host.CheckValue(args, nameof(args));
             Host.CheckUserArg(args.L2Weight >= 0, nameof(args.L2Weight), "L2 regularization term cannot be negative");