GetSummaryDataView() implementation for Pca and Linear Predictors

src/Microsoft.ML.PCA/PcaTrainer.cs

@@ -310,6 +310,7 @@ public static CommonOutputs.AnomalyDetectionOutput TrainPcaAnomaly(IHostEnvironm
     // REVIEW: move the predictor to a different file and fold EigenUtils.cs to this file.
     public sealed class PcaPredictor : PredictorBase<Float>,
         IValueMapper,
+        ICanGetSummaryAsIDataView,
         ICanSaveInTextFormat, ICanSaveModel, ICanSaveSummary
     {
         public const string LoaderSignature = "pcaAnomExec";
@@ -471,6 +472,16 @@ public void SaveAsText(TextWriter writer, RoleMappedSchema schema)
             }
         }
 
+        public IDataView GetSummaryDataView(RoleMappedSchema schema)
+        {
+            var bldr = new ArrayDataViewBuilder(Host);
+
+            bldr.AddColumn("MeanVector", NumberType.R4, _mean);
+            bldr.AddColumn("ProjectedMeanVector", NumberType.R4, _meanProjected);
+            bldr.AddColumn("EigenVectors", NumberType.R4, _eigenVectors);
+            return bldr.GetDataView();
+        }
+
         public ColumnType InputType
         {
             get { return _inputType; }

src/Microsoft.ML.StandardLearners/Standard/LinearPredictor.cs

@@ -44,6 +44,7 @@ public abstract class LinearPredictor : PredictorBase<Float>,
         ICanSaveInTextFormat,
         ICanSaveInSourceCode,
         ICanSaveModel,
+        ICanGetSummaryAsIRow,
         ICanSaveSummary,
         IPredictorWithFeatureWeights<Float>,
         IWhatTheFeatureValueMapper,
@@ -343,6 +344,30 @@ public void SaveAsCode(TextWriter writer, RoleMappedSchema schema)
 
         public abstract void SaveSummary(TextWriter writer, RoleMappedSchema schema);
 
+        public virtual IRow GetSummaryIRowOrNull(RoleMappedSchema schema)
+        {
+            var cols = new List<IColumn>();
+
+            var names = default(VBuffer<DvText>);
+            MetadataUtils.GetSlotNames(schema, RoleMappedSchema.ColumnRole.Feature, Weight.Length, ref names);
+            var slotNamesCol = RowColumnUtils.GetColumn(MetadataUtils.Kinds.SlotNames,
+                new VectorType(TextType.Instance, Weight.Length), ref names);
+            var slotNamesRow = RowColumnUtils.GetRow(null, slotNamesCol);
+            var colType = new VectorType(NumberType.R4, Weight.Length);
+
+            // Add the bias and the weight columns.
+            var bias = Bias;
+            cols.Add(RowColumnUtils.GetColumn("Bias", NumberType.R4, ref bias));
+            var weights = Weight;
+            cols.Add(RowColumnUtils.GetColumn("Weights", colType, ref weights, slotNamesRow));
+            return RowColumnUtils.GetRow(null, cols.ToArray());
+        }
+
+        public virtual IRow GetStatsIRowOrNull(RoleMappedSchema schema)
+        {
+            return null;
+        }
+
         public abstract void SaveAsIni(TextWriter writer, RoleMappedSchema schema, ICalibrator calibrator = null);
 
         public virtual void GetFeatureWeights(ref VBuffer<Float> weights)
@@ -366,8 +391,7 @@ public ValueMapper<TSrc, VBuffer<Float>> GetWhatTheFeatureMapper<TSrc, TDstContr
 
     public sealed partial class LinearBinaryPredictor : LinearPredictor,
         ICanGetSummaryInKeyValuePairs,
-        IParameterMixer<Float>,
-        ICanGetSummaryAsIRow
+        IParameterMixer<Float>
     {
         public const string LoaderSignature = "Linear2CExec";
         public const string RegistrationName = "LinearBinaryPredictor";
@@ -503,26 +527,7 @@ public IList<KeyValuePair<string, object>> GetSummaryInKeyValuePairs(RoleMappedS
             return results;
         }
 
-        public IRow GetSummaryIRowOrNull(RoleMappedSchema schema)
-        {
-            var cols = new List<IColumn>();
-
-            var names = default(VBuffer<DvText>);
-            MetadataUtils.GetSlotNames(schema, RoleMappedSchema.ColumnRole.Feature, Weight.Length, ref names);
-            var slotNamesCol = RowColumnUtils.GetColumn(MetadataUtils.Kinds.SlotNames,
-                new VectorType(TextType.Instance, Weight.Length), ref names);
-            var slotNamesRow = RowColumnUtils.GetRow(null, slotNamesCol);
-            var colType = new VectorType(NumberType.R4, Weight.Length);
-
-            // Add the bias and the weight columns.
-            var bias = Bias;
-            cols.Add(RowColumnUtils.GetColumn("Bias", NumberType.R4, ref bias));
-            var weights = Weight;
-            cols.Add(RowColumnUtils.GetColumn("Weights", colType, ref weights, slotNamesRow));
-            return RowColumnUtils.GetRow(null, cols.ToArray());
-        }
-
-        public IRow GetStatsIRowOrNull(RoleMappedSchema schema)
+        public override IRow GetStatsIRowOrNull(RoleMappedSchema schema)
         {
             if (_stats == null)
                 return null;
@@ -582,8 +587,7 @@ public override void SaveAsIni(TextWriter writer, RoleMappedSchema schema, ICali
 
     public sealed class LinearRegressionPredictor : RegressionPredictor,
         IParameterMixer<Float>,
-        ICanGetSummaryInKeyValuePairs,
-        ICanGetSummaryAsIRow
+        ICanGetSummaryInKeyValuePairs
     {
         public const string LoaderSignature = "LinearRegressionExec";
         public const string RegistrationName = "LinearRegressionPredictor";
@@ -663,30 +667,6 @@ public IList<KeyValuePair<string, object>> GetSummaryInKeyValuePairs(RoleMappedS
 
             return results;
         }
-
-        public IRow GetSummaryIRowOrNull(RoleMappedSchema schema)
-        {
-            var cols = new List<IColumn>();
-
-            var names = default(VBuffer<DvText>);
-            MetadataUtils.GetSlotNames(schema, RoleMappedSchema.ColumnRole.Feature, Weight.Length, ref names);
-            var slotNamesCol = RowColumnUtils.GetColumn(MetadataUtils.Kinds.SlotNames,
-                new VectorType(TextType.Instance, Weight.Length), ref names);
-            var slotNamesRow = RowColumnUtils.GetRow(null, slotNamesCol);
-            var colType = new VectorType(NumberType.R4, Weight.Length);
-
-            // Add the bias and the weight columns.
-            var bias = Bias;
-            cols.Add(RowColumnUtils.GetColumn("Bias", NumberType.R4, ref bias));
-            var weights = Weight;
-            cols.Add(RowColumnUtils.GetColumn("Weights", colType, ref weights, slotNamesRow));
-            return RowColumnUtils.GetRow(null, cols.ToArray());
-        }
-
-        public IRow GetStatsIRowOrNull(RoleMappedSchema schema)
-        {
-            return null;
-        }
     }
 
     public sealed class PoissonRegressionPredictor : RegressionPredictor, IParameterMixer<Float>

test/BaselineOutput/SingleDebug/pcaAnomaly/Default-TrainTest-azureCounterUnlabeled-out.txt

@@ -0,0 +1,24 @@
+maml.exe TrainTest test=%Data%\azure-test-unlabeled.txt tr=pcaAnomaly dout=%Output%\pcaAnomaly\Default-TrainTest-azureCounterUnlabeled.txt loader=Text{sep=space col=Name:TX:0 col=Features:~} data=%Data%\azure-train-unlabeled.txt out=%Output%\pcaAnomaly\Default-TrainTest-azureCounterUnlabeled-model.zip seed=1 xf=Expr{col=Label:Name expr={x : na(float(0))}}
+Automatically adding a MinMax normalization transform, use 'norm=Warn' or 'norm=No' to turn this behavior off.
+Not training a calibrator because it is not needed.
+Warning: Encountered 6590 unlabeled instances during testing.
+Detection rate at 10 false positives:     NaN
+Detection rate at 0.01 false positive rate:     NaN
+Detection rate at 0 positive predictions:     NaN
+Threshold @K FP:        NaN
+Threshold @P FPR:       NaN
+Threshold @NumPos:      NaN
+AUC:                    NaN
+
+OVERALL RESULTS
+---------------------------------------
+AUC:                    NaN (0.0000)
+DR @K FP:               NaN (0.0000)
+DR @P FPR:              NaN (0.0000)
+DR @NumPos:             NaN (0.0000)
+
+---------------------------------------
+Physical memory usage(MB): %Number%
+Virtual memory usage(MB): %Number%
+%DateTime%	 Time elapsed(s): %Number%
+

test/BaselineOutput/SingleDebug/pcaAnomaly/Default-TrainTest-azureCounterUnlabeled-rp.txt

@@ -0,0 +1,4 @@
+pcaAnomaly
+AUC	DR @K FP	DR @P FPR	DR @NumPos	Learner Name	Train Dataset	Test Dataset	Results File	Run Time	Physical Memory	Virtual Memory	Command Line	Settings	
+NaN	NaN	NaN	NaN	pcaAnomaly	%Data%\azure-train-unlabeled.txt	%Data%\azure-test-unlabeled.txt	%Output%\pcaAnomaly\Default-TrainTest-azureCounterUnlabeled-out.txt	99	0	0	maml.exe TrainTest test=%Data%\azure-test-unlabeled.txt tr=pcaAnomaly dout=%Output%\pcaAnomaly\Default-TrainTest-azureCounterUnlabeled.txt loader=Text{sep=space col=Name:TX:0 col=Features:~} data=%Data%\azure-train-unlabeled.txt out=%Output%\pcaAnomaly\Default-TrainTest-azureCounterUnlabeled-model.zip seed=1 xf=Expr{col=Label:Name expr={x : na(float(0))}}		
+