update tests to check categorical splits

Author: artidoro

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

@@ -177,6 +177,18 @@ public void TestEstimatorMulticlassNaiveBayesTrainer()
             return (pipeline, data);
         }
 
+        /// <summary>
+        /// Same data as <see cref="GetBinaryClassificationPipeline"/>, but with additional
+        /// OneHotEncoding to obtain categorical splits in tree models.
+        /// </summary>
+        private (IEstimator<ITransformer>, IDataView) GetOneHotBinaryClassificationPipeline()
+        {
+            var (pipeline, data) = GetBinaryClassificationPipeline();
+            var oneHotPipeline = pipeline.Append(ML.Transforms.Categorical.OneHotEncoding("Features"));
+
+            return (oneHotPipeline, data);
+        }
+
 
         private (IEstimator<ITransformer>, IDataView) GetRankingPipeline()
         {

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

@@ -685,9 +685,9 @@ private void CheckSummary(ICanGetSummaryAsIDataView modelParameters, double bias
             IEnumerable<SummaryDataRow> summaryDataEnumerable;
 
             if (quantileTrees == null)
-                summaryDataEnumerable = ML.Data.CreateEnumerable<SummaryDataRow>(summaryDataView, true);
+                summaryDataEnumerable = ML.Data.CreateEnumerable<SummaryDataRow>(summaryDataView, false);
             else
-                summaryDataEnumerable = ML.Data.CreateEnumerable<QuantileTestSummaryDataRow>(summaryDataView, true);
+                summaryDataEnumerable = ML.Data.CreateEnumerable<QuantileTestSummaryDataRow>(summaryDataView, false);
 
             var summaryDataEnumerator = summaryDataEnumerable.GetEnumerator();
 
@@ -720,7 +720,7 @@ public void FastTreeRegressorTestSummary()
         {
             var dataView = GetRegressionPipeline();
             var trainer = ML.Regression.Trainers.FastTree(
-                new FastTreeRegressionTrainer.Options { NumberOfTrees = 10, NumberOfThreads = 1, NumberOfLeaves = 5, CategoricalSplit = true});
+                new FastTreeRegressionTrainer.Options { NumberOfTrees = 10, NumberOfThreads = 1, NumberOfLeaves = 5});
 
             var transformer = trainer.Fit(dataView);
 
@@ -738,7 +738,7 @@ public void FastForestRegressorTestSummary()
         {
             var dataView = GetRegressionPipeline();
             var trainer = ML.Regression.Trainers.FastForest(
-                new FastForestRegressionTrainer.Options { NumberOfTrees = 10, NumberOfThreads = 1, NumberOfLeaves = 5, CategoricalSplit = true });
+                new FastForestRegressionTrainer.Options { NumberOfTrees = 10, NumberOfThreads = 1, NumberOfLeaves = 5});
 
             var transformer = trainer.Fit(dataView);
 
@@ -756,7 +756,7 @@ public void FastTreeTweedieRegressorTestSummary()
         {
             var dataView = GetRegressionPipeline();
             var trainer = ML.Regression.Trainers.FastTreeTweedie(
-                new FastTreeTweedieTrainer.Options { NumberOfTrees = 10, NumberOfThreads = 1, NumberOfLeaves = 5, CategoricalSplit = true });
+                new FastTreeTweedieTrainer.Options { NumberOfTrees = 10, NumberOfThreads = 1, NumberOfLeaves = 5});
 
             var transformer = trainer.Fit(dataView);
 
@@ -774,7 +774,7 @@ public void LightGbmRegressorTestSummary()
         {
             var dataView = GetRegressionPipeline();
             var trainer = ML.Regression.Trainers.LightGbm(
-                new LightGbmRegressionTrainer.Options { NumberOfIterations = 10, NumberOfThreads = 1, NumberOfLeaves = 5, UseCategoricalSplit = true });
+                new LightGbmRegressionTrainer.Options { NumberOfIterations = 10, NumberOfThreads = 1, NumberOfLeaves = 5});
 
             var transformer = trainer.Fit(dataView);
 
@@ -792,7 +792,7 @@ public void FastTreeBinaryClassificationTestSummary()
         {
             var (pipeline, dataView) = GetBinaryClassificationPipeline();
             var estimator = pipeline.Append(ML.BinaryClassification.Trainers.FastTree(
-                new FastTreeBinaryTrainer.Options { NumberOfTrees = 10, NumberOfThreads = 1, NumberOfLeaves = 5, CategoricalSplit = true }));
+                new FastTreeBinaryTrainer.Options { NumberOfTrees = 2, NumberOfThreads = 1, NumberOfLeaves = 5}));
 
             var transformer = estimator.Fit(dataView);
 
@@ -808,9 +808,9 @@ public void FastTreeBinaryClassificationTestSummary()
         [Fact]
         public void FastForestBinaryClassificationTestSummary()
         {
-            var (pipeline, dataView) = GetBinaryClassificationPipeline();
+            var (pipeline, dataView) = GetOneHotBinaryClassificationPipeline();
             var estimator = pipeline.Append(ML.BinaryClassification.Trainers.FastForest(
-                new FastForestBinaryTrainer.Options { NumberOfTrees = 10, NumberOfThreads = 1, NumberOfLeaves = 5, CategoricalSplit = true }));
+                new FastForestBinaryTrainer.Options { NumberOfTrees = 2, NumberOfThreads = 1, NumberOfLeaves = 4, CategoricalSplit = true }));
 
             var transformer = estimator.Fit(dataView);
 
@@ -826,7 +826,7 @@ public void FastForestBinaryClassificationTestSummary()
         [LightGBMFact]
         public void LightGbmBinaryClassificationTestSummary()
         {
-            var (pipeline, dataView) = GetBinaryClassificationPipeline();
+            var (pipeline, dataView) = GetOneHotBinaryClassificationPipeline();
             var trainer = pipeline.Append(ML.BinaryClassification.Trainers.LightGbm(
                 new LightGbmBinaryTrainer.Options { NumberOfIterations = 10, NumberOfThreads = 1, NumberOfLeaves = 5, UseCategoricalSplit = true }));