Conversion of DropSlots, MutualInformationFeatureSelection, and CountFeatureSelection into estimator and transformers

docs/code/MlNetCookBook.md

@@ -1055,7 +1055,7 @@ var dynamicPipeline =
     .Append(mlContext.Transforms.Categorical.OneHotEncoding("CategoricalFeatures", "CategoricalBag", CategoricalTransform.OutputKind.Bag))
     // One-hot encode the workclass column, then drop all the categories that have fewer than 10 instances in the train set.
     .Append(mlContext.Transforms.Categorical.OneHotEncoding("Workclass", "WorkclassOneHot"))
-    .Append(new CountFeatureSelector(mlContext, "WorkclassOneHot", "WorkclassOneHotTrimmed", count: 10));
+    .Append(mlContext.Transforms.FeatureSelection.CountFeatureSelectingEstimator("WorkclassOneHot", "WorkclassOneHotTrimmed", count: 10));
 
 // Let's train our pipeline, and then apply it to the same data.
 var transformedData = dynamicPipeline.Fit(data).Transform(data);

docs/samples/Microsoft.ML.Samples/Dynamic/FeatureSelectionTransform.cs

@@ -0,0 +1,123 @@
+using Microsoft.ML.Data;
+using Microsoft.ML.Runtime.Data;
+using System;
+using System.Collections.Generic;
+
+namespace Microsoft.ML.Samples.Dynamic
+{
+    public class FeatureSelectionTransformExample
+    {
+        public static void FeatureSelectionTransform()
+        {
+            // Downloading a classification dataset from github.com/dotnet/machinelearning.
+            // It will be stored in the same path as the executable
+            string dataFilePath = SamplesUtils.DatasetUtils.DownloadBreastCancerDataset();
+
+            // Data Preview
+            //    1. Label							0=benign, 1=malignant
+            //    2. Clump Thickness               1 - 10
+            //    3. Uniformity of Cell Size       1 - 10
+            //    4. Uniformity of Cell Shape      1 - 10
+            //    5. Marginal Adhesion             1 - 10
+            //    6. Single Epithelial Cell Size   1 - 10
+            //    7. Bare Nuclei                   1 - 10
+            //    8. Bland Chromatin               1 - 10
+            //    9. Normal Nucleoli               1 - 10
+            //   10. Mitoses                       1 - 10
+
+            // Create a new ML context, for ML.NET operations. It can be used for exception tracking and logging, 
+            // as well as the source of randomness.
+            var ml = new MLContext();
+
+            // First, we define the reader: specify the data columns and where to find them in the text file. Notice that we combine entries from
+            // all the feature columns into entries of a vector of a single column named "Features".
+            var reader = ml.Data.TextReader(new TextLoader.Arguments()
+            {
+                Separator = "tab",
+                HasHeader = true,
+                Column = new[]
+                    {
+                        new TextLoader.Column("Label", DataKind.BL, 0),
+                        new TextLoader.Column("Features", DataKind.Num, new [] { new TextLoader.Range(1, 9) })
+                    }
+            });
+
+            // Then, we use the reader to read the data as an IDataView.
+            var data = reader.Read(dataFilePath);
+
+            // Second, we define the transformations that we apply on the data. Remember that an Estimator does not transform data
+            // directly, but it needs to be trained on data using .Fit(), and it will output a Transformer, which can transform data.
+
+            // In this example we define a CountFeatureSelectingEstimator, that selects slots in a feature vector that have more non-default 
+            // values than the specified count. This transformation can be used to remove slots with too many missing values.
+            var countSelectEst = ml.Transforms.FeatureSelection.SelectFeaturesBasedOnCount(
+                inputColumn: "Features", outputColumn: "FeaturesCountSelect", count: 695);
+
+            // We also define a MutualInformationFeatureSelectingEstimator that selects the top k slots in a feature 
+            // vector based on highest mutual information between that slot and a specified label. Notice that it is possible to 
+            // specify the parameter `numBins', which controls the number of bins used in the approximation of the mutual information
+            // between features and label.
+            var mutualInfoEst = ml.Transforms.FeatureSelection.SelectFeaturesBasedOnMutualInformation(
+                inputColumn: "FeaturesCountSelect", outputColumn: "FeaturesMISelect", labelColumn: "Label", slotsInOutput: 5);
+
+            // Now, we can put the previous two transformations together in a pipeline.
+            var pipeline = countSelectEst.Append(mutualInfoEst);
+
+            // The pipeline can then be trained, using .Fit(), and the resulting transformer can be used to transform data. 
+            var transformedData = pipeline.Fit(data).Transform(data);
+
+            // Small helper to print the data inside a column, in the console. Only prints the first 10 rows.
+            Action<string, IEnumerable<VBuffer<float>>> printHelper = (columnName, column) =>
+            {
+                Console.WriteLine($"{columnName} column obtained post-transformation.");
+                int count = 0;
+                foreach (var row in column)
+                {
+                    foreach (var value in row.GetValues())
+                        Console.Write($"{value}\t");
+                    Console.WriteLine("");
+                    count++;
+                    if (count >= 10)
+                        break;
+                }
+
+                Console.WriteLine("===================================================");
+            };
+
+            // Print the data that results from the transformations.
+            var countSelectColumn = transformedData.GetColumn<VBuffer<float>>(ml, "FeaturesCountSelect");
+            var MISelectColumn = transformedData.GetColumn<VBuffer<float>>(ml, "FeaturesMISelect");
+            printHelper("FeaturesCountSelect", countSelectColumn);
+            printHelper("FeaturesMISelect", MISelectColumn);
+
+            // Below is the output of the this code. We see that some slots habe been dropped by the first transformation.
+            // Among the remaining slots, the second transformation only preserves the top 5 slots based on mutualinformation 
+            // with the label column.
+
+            // FeaturesCountSelect column obtained post-transformation.
+            // 5       4       4       5       7       3       2       1
+            // 3       1       1       1       2       3       1       1
+            // 6       8       8       1       3       3       7       1
+            // 4       1       1       3       2       3       1       1
+            // 8       10      10      8       7       9       7       1
+            // 1       1       1       1       2       3       1       1
+            // 2       1       2       1       2       3       1       1
+            // 2       1       1       1       2       1       1       5
+            // 4       2       1       1       2       2       1       1
+            // 1       1       1       1       1       3       1       1
+            // ===================================================
+            // FeaturesMISelect column obtained post-transformation.
+            // 4       4       7       3       2
+            // 1       1       2       3       1
+            // 8       8       3       3       7
+            // 1       1       2       3       1
+            // 10      10      7       9       7
+            // 1       1       2       3       1
+            // 1       2       2       3       1
+            // 1       1       2       1       1
+            // 2       1       2       2       1
+            // 1       1       1       3       1
+            // ===================================================
+        }
+    }
+}

docs/samples/Microsoft.ML.Samples/Static/FastTreeBinaryClassification.cs

@@ -2,6 +2,7 @@
 using Microsoft.ML.StaticPipe;
 using Microsoft.ML.Transforms;
 using Microsoft.ML.Transforms.Categorical;
+using Microsoft.ML.Transforms.FeatureSelection;
 using System;
 
 namespace Microsoft.ML.Samples.Static

docs/samples/Microsoft.ML.Samples/Static/FeatureSelectionTransform.cs

@@ -0,0 +1,122 @@
+using Microsoft.ML.Data;
+using Microsoft.ML.Runtime.Data;
+using Microsoft.ML.StaticPipe;
+using System;
+using System.Collections.Generic;
+
+namespace Microsoft.ML.Samples.Dynamic
+{
+    public class FeatureSelectionTransformStaticExample
+    {
+        public static void FeatureSelectionTransform()
+        {
+            // Downloading a classification dataset from github.com/dotnet/machinelearning.
+            // It will be stored in the same path as the executable
+            string dataFilePath = SamplesUtils.DatasetUtils.DownloadBreastCancerDataset();
+
+            // Data Preview
+            //    1. Label							0=benign, 1=malignant
+            //    2. Clump Thickness               1 - 10
+            //    3. Uniformity of Cell Size       1 - 10
+            //    4. Uniformity of Cell Shape      1 - 10
+            //    5. Marginal Adhesion             1 - 10
+            //    6. Single Epithelial Cell Size   1 - 10
+            //    7. Bare Nuclei                   1 - 10
+            //    8. Bland Chromatin               1 - 10
+            //    9. Normal Nucleoli               1 - 10
+            //   10. Mitoses                       1 - 10
+
+            // Create a new ML context, for ML.NET operations. It can be used for exception tracking and logging, 
+            // as well as the source of randomness.
+            var ml = new MLContext();
+
+            // First, we define the reader: specify the data columns and where to find them in the text file. Notice that we combine entries from
+            // all the feature columns into entries of a vector of a single column named "Features".
+            var reader = TextLoader.CreateReader(ml, c => (
+                        Label: c.LoadBool(0),
+                        Features: c.LoadFloat(1, 9)
+                    ),
+                separator: '\t', hasHeader: true);
+
+            // Then, we use the reader to read the data as an IDataView.
+            var data = reader.Read(dataFilePath);
+
+            // Second, we define the transformations that we apply on the data. Remember that an Estimator does not transform data
+            // directly, but it needs to be trained on data using .Fit(), and it will output a Transformer, which can transform data.
+
+            // In this example we define a CountFeatureSelectingEstimator, that selects slots in a feature vector that have more non-default 
+            // values than the specified count. This transformation can be used to remove slots with too many missing values.
+            // We also define a MutualInformationFeatureSelectingEstimator that selects the top k slots in a feature 
+            // vector based on highest mutual information between that slot and a specified label. Notice that it is possible to 
+            // specify the parameter `numBins', which controls the number of bins used in the approximation of the mutual information
+            // between features and label.
+            var pipeline = reader.MakeNewEstimator()
+                .Append(r =>(
+                    FeaturesCountSelect: r.Features.SelectFeaturesBasedOnCount(count: 695),
+                    Label: r.Label
+                    ))
+                .Append(r => (
+                    FeaturesCountSelect: r.FeaturesCountSelect,
+                    FeaturesMISelect: r.FeaturesCountSelect.SelectFeaturesBasedOnMutualInformation(r.Label, slotsInOutput: 5),
+                    Label: r.Label
+                    ));
+
+
+            // The pipeline can then be trained, using .Fit(), and the resulting transformer can be used to transform data. 
+            var transformedData = pipeline.Fit(data).Transform(data);
+
+            // Small helper to print the data inside a column, in the console. Only prints the first 10 rows.
+            Action<string, IEnumerable<VBuffer<float>>> printHelper = (columnName, column) =>
+            {
+                Console.WriteLine($"{columnName} column obtained post-transformation.");
+                int count = 0;
+                foreach (var row in column)
+                {
+                    foreach (var value in row.GetValues())
+                        Console.Write($"{value}\t");
+                    Console.WriteLine("");
+                    count++;
+                    if (count >= 10)
+                        break;
+                }
+
+                Console.WriteLine("===================================================");
+            };
+
+            // Print the data that results from the transformations.
+            var countSelectColumn = transformedData.AsDynamic.GetColumn<VBuffer<float>>(ml, "FeaturesCountSelect");
+            var MISelectColumn = transformedData.AsDynamic.GetColumn<VBuffer<float>>(ml, "FeaturesMISelect");
+            printHelper("FeaturesCountSelect", countSelectColumn);
+            printHelper("FeaturesMISelect", MISelectColumn);
+
+            // Below is the output of the this code. We see that some slots habe been dropped by the first transformation.
+            // Among the remaining slots, the second transformation only preserves the top 5 slots based on mutualinformation 
+            // with the label column.
+
+            // FeaturesCountSelect column obtained post-transformation.
+            // 5       4       4       5       7       3       2       1
+            // 3       1       1       1       2       3       1       1
+            // 6       8       8       1       3       3       7       1
+            // 4       1       1       3       2       3       1       1
+            // 8       10      10      8       7       9       7       1
+            // 1       1       1       1       2       3       1       1
+            // 2       1       2       1       2       3       1       1
+            // 2       1       1       1       2       1       1       5
+            // 4       2       1       1       2       2       1       1
+            // 1       1       1       1       1       3       1       1
+            // ===================================================
+            // FeaturesMISelect column obtained post-transformation.
+            // 4       4       7       3       2
+            // 1       1       2       3       1
+            // 8       8       3       3       7
+            // 1       1       2       3       1
+            // 10      10      7       9       7
+            // 1       1       2       3       1
+            // 1       2       2       3       1
+            // 1       1       2       1       1
+            // 2       1       2       2       1
+            // 1       1       1       3       1
+            // ===================================================
+        }
+    }
+}

docs/samples/Microsoft.ML.Samples/Static/LightGBMBinaryClassification.cs

@@ -2,6 +2,7 @@
 using Microsoft.ML.StaticPipe;
 using Microsoft.ML.Transforms;
 using Microsoft.ML.Transforms.Categorical;
+using Microsoft.ML.Transforms.FeatureSelection;
 using System;
 
 namespace Microsoft.ML.Samples.Static

docs/samples/Microsoft.ML.Samples/Static/SDCABinaryClassification.cs

@@ -2,6 +2,7 @@
 using Microsoft.ML.StaticPipe;
 using Microsoft.ML.Transforms;
 using Microsoft.ML.Transforms.Categorical;
+using Microsoft.ML.Transforms.FeatureSelection;
 using System;
 
 namespace Microsoft.ML.Samples.Static

src/Microsoft.ML.Data/Evaluators/ClusteringEvaluator.cs

@@ -10,7 +10,7 @@
 using Microsoft.ML.Runtime.Internal.Utilities;
 using Microsoft.ML.Runtime.Model;
 using Microsoft.ML.Runtime.Numeric;
-using Microsoft.ML.Transforms;
+using Microsoft.ML.Transforms.FeatureSelection;
 using System;
 using System.Collections.Generic;
 using System.Linq;
@@ -877,50 +877,14 @@ protected override IDataView GetPerInstanceMetricsCore(IDataView perInst, RoleMa
             {
                 var type = perInst.Schema.GetColumnType(index);
                 if (_numTopClusters < type.VectorSize)
-                {
-                    var args = new DropSlotsTransform.Arguments
-                    {
-                        Column = new DropSlotsTransform.Column[]
-                        {
-                            new DropSlotsTransform.Column()
-                            {
-                                Name = ClusteringPerInstanceEvaluator.SortedClusters,
-                                Slots = new[] {
-                                    new DropSlotsTransform.Range()
-                                    {
-                                        Min = _numTopClusters
-                                    }
-                                }
-                            }
-                        }
-                    };
-                    perInst = new DropSlotsTransform(Host, args, perInst);
-                }
+                    perInst = new SlotsDroppingTransformer(Host, ClusteringPerInstanceEvaluator.SortedClusters, min: _numTopClusters).Transform(perInst);
             }
 
             if (perInst.Schema.TryGetColumnIndex(ClusteringPerInstanceEvaluator.SortedClusterScores, out index))
             {
                 var type = perInst.Schema.GetColumnType(index);
                 if (_numTopClusters < type.VectorSize)
-                {
-                    var args = new DropSlotsTransform.Arguments
-                    {
-                        Column = new DropSlotsTransform.Column[]
-                        {
-                            new DropSlotsTransform.Column()
-                            {
-                                Name = ClusteringPerInstanceEvaluator.SortedClusterScores,
-                                Slots = new[] {
-                                    new DropSlotsTransform.Range()
-                                    {
-                                        Min = _numTopClusters
-                                    }
-                                }
-                            }
-                        }
-                    };
-                    perInst = new DropSlotsTransform(Host, args, perInst);
-                }
+                    perInst = new SlotsDroppingTransformer(Host, ClusteringPerInstanceEvaluator.SortedClusterScores, min: _numTopClusters).Transform(perInst);
             }
             return perInst;
         }