Have EvaluatorUtils and calling code exploit Schema.Column not ColumnInfo.

Author: TomFinley

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

@@ -749,9 +749,9 @@ private protected override IEnumerable<string> GetPerInstanceColumnsToSave(RoleM
 
             // The anomaly detection evaluator outputs the label and the score.
             yield return schema.Label.Value.Name;
-            var scoreInfo = EvaluateUtils.GetScoreColumnInfo(Host, schema.Schema, ScoreCol, nameof(Arguments.ScoreColumn),
+            var scoreCol = EvaluateUtils.GetScoreColumn(Host, schema.Schema, ScoreCol, nameof(Arguments.ScoreColumn),
                 MetadataUtils.Const.ScoreColumnKind.AnomalyDetection);
-            yield return scoreInfo.Name;
+            yield return scoreCol.Name;
 
             // No additional output columns.
         }

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

@@ -1175,14 +1175,14 @@ public BinaryClassifierMamlEvaluator(IHostEnvironment env, Arguments args)
         {
             var cols = base.GetInputColumnRolesCore(schema);
 
-            var scoreInfo = EvaluateUtils.GetScoreColumnInfo(Host, schema.Schema, ScoreCol, nameof(Arguments.ScoreColumn),
+            var scoreCol = EvaluateUtils.GetScoreColumn(Host, schema.Schema, ScoreCol, nameof(Arguments.ScoreColumn),
                 MetadataUtils.Const.ScoreColumnKind.BinaryClassification);
 
             // Get the optional probability column.
-            var probInfo = EvaluateUtils.GetOptAuxScoreColumnInfo(Host, schema.Schema, _probCol, nameof(Arguments.ProbabilityColumn),
-                scoreInfo.Index, MetadataUtils.Const.ScoreValueKind.Probability, t => t == NumberType.Float);
-            if (probInfo != null)
-                cols = MetadataUtils.Prepend(cols, RoleMappedSchema.CreatePair(MetadataUtils.Const.ScoreValueKind.Probability, probInfo.Name));
+            var probCol = EvaluateUtils.GetOptAuxScoreColumn(Host, schema.Schema, _probCol, nameof(Arguments.ProbabilityColumn),
+                scoreCol.Index, MetadataUtils.Const.ScoreValueKind.Probability, NumberType.Float.Equals);
+            if (probCol.HasValue)
+                cols = MetadataUtils.Prepend(cols, RoleMappedSchema.CreatePair(MetadataUtils.Const.ScoreValueKind.Probability, probCol.Value.Name));
             return cols;
         }
 
@@ -1485,15 +1485,15 @@ private protected override IEnumerable<string> GetPerInstanceColumnsToSave(RoleM
 
             // The binary classifier evaluator outputs the label, score and probability columns.
             yield return schema.Label.Value.Name;
-            var scoreInfo = EvaluateUtils.GetScoreColumnInfo(Host, schema.Schema, ScoreCol, nameof(Arguments.ScoreColumn),
+            var scoreCol = EvaluateUtils.GetScoreColumn(Host, schema.Schema, ScoreCol, nameof(Arguments.ScoreColumn),
                 MetadataUtils.Const.ScoreColumnKind.BinaryClassification);
-            yield return scoreInfo.Name;
-            var probInfo = EvaluateUtils.GetOptAuxScoreColumnInfo(Host, schema.Schema, _probCol, nameof(Arguments.ProbabilityColumn),
-                scoreInfo.Index, MetadataUtils.Const.ScoreValueKind.Probability, t => t == NumberType.Float);
+            yield return scoreCol.Name;
+            var probCol = EvaluateUtils.GetOptAuxScoreColumn(Host, schema.Schema, _probCol, nameof(Arguments.ProbabilityColumn),
+                scoreCol.Index, MetadataUtils.Const.ScoreValueKind.Probability, NumberType.Float.Equals);
             // Return the output columns. The LogLoss column is returned only if the probability column exists.
-            if (probInfo != null)
+            if (probCol.HasValue)
             {
-                yield return probInfo.Name;
+                yield return probCol.Value.Name;
                 yield return BinaryPerInstanceEvaluator.LogLoss;
             }
 

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

@@ -101,11 +101,11 @@ private static bool CheckScoreColumnKind(Schema schema, int col)
         }
 
         /// <summary>
-        /// Find the score column to use. If name is specified, that is used. Otherwise, this searches for the
-        /// most recent score set of the given kind. If there is no such score set and defName is specifed it
-        /// uses defName. Otherwise, it throws.
+        /// Find the score column to use. If <paramref name="name"/> is specified, that is used. Otherwise, this searches
+        /// for the most recent score set of the given <paramref name="kind"/>. If there is no such score set and
+        /// <paramref name="defName"/> is specifed it uses <paramref name="defName"/>. Otherwise, it throws.
         /// </summary>
-        public static ColumnInfo GetScoreColumnInfo(IExceptionContext ectx, Schema schema, string name, string argName, string kind,
+        public static Schema.Column GetScoreColumn(IExceptionContext ectx, Schema schema, string name, string argName, string kind,
             string valueKind = MetadataUtils.Const.ScoreValueKind.Score, string defName = null)
         {
             Contracts.CheckValueOrNull(ectx);
@@ -115,51 +115,52 @@ public static ColumnInfo GetScoreColumnInfo(IExceptionContext ectx, Schema schem
             ectx.CheckNonEmpty(kind, nameof(kind));
             ectx.CheckNonEmpty(valueKind, nameof(valueKind));
 
-            int colTmp;
-            ColumnInfo info;
             if (!string.IsNullOrWhiteSpace(name))
             {
-#pragma warning disable MSML_ContractsNameUsesNameof
-                if (!ColumnInfo.TryCreateFromName(schema, name, out info))
+#pragma warning disable MSML_ContractsNameUsesNameof // This utility method is meant to reflect the argument name of whatever is calling it, so we take that as a parameter, rather than using nameof directly as in most cases.
+                var col = schema.GetColumnOrNull(name);
+                if (!col.HasValue)
                     throw ectx.ExceptUserArg(argName, "Score column is missing");
 #pragma warning restore MSML_ContractsNameUsesNameof
-                return info;
+                return col.Value;
             }
 
-            var maxSetNum = schema.GetMaxMetadataKind(out colTmp, MetadataUtils.Kinds.ScoreColumnSetId,
+            var maxSetNum = schema.GetMaxMetadataKind(out int colTmp, MetadataUtils.Kinds.ScoreColumnSetId,
                 (s, c) => IsScoreColumnKind(ectx, s, c, kind));
 
             ReadOnlyMemory<char> tmp = default;
-            foreach (var col in schema.GetColumnSet(MetadataUtils.Kinds.ScoreColumnSetId, maxSetNum))
+            foreach (var colIdx in schema.GetColumnSet(MetadataUtils.Kinds.ScoreColumnSetId, maxSetNum))
             {
+                var col = schema[colIdx];
 #if DEBUG
-                schema[col].Metadata.GetValue(MetadataUtils.Kinds.ScoreColumnKind, ref tmp);
+                col.Metadata.GetValue(MetadataUtils.Kinds.ScoreColumnKind, ref tmp);
                 ectx.Assert(ReadOnlyMemoryUtils.EqualsStr(kind, tmp));
 #endif
                 // REVIEW: What should this do about hidden columns? Currently we ignore them.
-                if (schema[col].IsHidden)
+                if (col.IsHidden)
                     continue;
-                if (schema.TryGetMetadata(TextType.Instance, MetadataUtils.Kinds.ScoreValueKind, col, ref tmp) &&
-                    ReadOnlyMemoryUtils.EqualsStr(valueKind, tmp))
+                if (col.Metadata.Schema.GetColumnOrNull(MetadataUtils.Kinds.ScoreValueKind)?.Type == TextType.Instance)
                 {
-                    return ColumnInfo.CreateFromIndex(schema, col);
+                    col.Metadata.GetValue(MetadataUtils.Kinds.ScoreValueKind, ref tmp);
+                    if (ReadOnlyMemoryUtils.EqualsStr(valueKind, tmp))
+                        return col;
                 }
             }
 
-            if (!string.IsNullOrWhiteSpace(defName) && ColumnInfo.TryCreateFromName(schema, defName, out info))
-                return info;
+            if (!string.IsNullOrWhiteSpace(defName) && schema.GetColumnOrNull(defName) is Schema.Column defCol)
+                return defCol;
 
 #pragma warning disable MSML_ContractsNameUsesNameof
             throw ectx.ExceptUserArg(argName, "Score column is missing");
 #pragma warning restore MSML_ContractsNameUsesNameof
         }
 
         /// <summary>
-        /// Find the optional auxilliary score column to use. If name is specified, that is used.
-        /// Otherwise, if colScore is part of a score set, this looks in the score set for a column
-        /// with the given valueKind. If none is found, it returns null.
+        /// Find the optional auxilliary score column to use. If <paramref name="name"/> is specified, that is used.
+        /// Otherwise, if <paramref name="colScore"/> is part of a score set, this looks in the score set for a column
+        /// with the given <paramref name="valueKind"/>. If none is found, it returns <see langword="null"/>.
         /// </summary>
-        public static ColumnInfo GetOptAuxScoreColumnInfo(IExceptionContext ectx, Schema schema, string name, string argName,
+        public static Schema.Column? GetOptAuxScoreColumn(IExceptionContext ectx, Schema schema, string name, string argName,
             int colScore, string valueKind, Func<ColumnType, bool> testType)
         {
             Contracts.CheckValueOrNull(ectx);
@@ -171,14 +172,14 @@ public static ColumnInfo GetOptAuxScoreColumnInfo(IExceptionContext ectx, Schema
 
             if (!string.IsNullOrWhiteSpace(name))
             {
-                ColumnInfo info;
 #pragma warning disable MSML_ContractsNameUsesNameof
-                if (!ColumnInfo.TryCreateFromName(schema, name, out info))
+                var col = schema.GetColumnOrNull(name);
+                if (!col.HasValue)
                     throw ectx.ExceptUserArg(argName, "{0} column is missing", valueKind);
-                if (!testType(info.Type))
+                if (!testType(col.Value.Type))
                     throw ectx.ExceptUserArg(argName, "{0} column has incompatible type", valueKind);
 #pragma warning restore MSML_ContractsNameUsesNameof
-                return info;
+                return col.Value;
             }
 
             // Get the score column set id from colScore.
@@ -192,17 +193,18 @@ public static ColumnInfo GetOptAuxScoreColumnInfo(IExceptionContext ectx, Schema
             schema[colScore].Metadata.GetValue(MetadataUtils.Kinds.ScoreColumnSetId, ref setId);
 
             ReadOnlyMemory<char> tmp = default;
-            foreach (var col in schema.GetColumnSet(MetadataUtils.Kinds.ScoreColumnSetId, setId))
+            foreach (var colIdx in schema.GetColumnSet(MetadataUtils.Kinds.ScoreColumnSetId, setId))
             {
                 // REVIEW: What should this do about hidden columns? Currently we ignore them.
-                if (schema[col].IsHidden)
+                var col = schema[colIdx];
+                if (col.IsHidden)
                     continue;
-                if (schema.TryGetMetadata(TextType.Instance, MetadataUtils.Kinds.ScoreValueKind, col, ref tmp) &&
-                    ReadOnlyMemoryUtils.EqualsStr(valueKind, tmp))
+
+                if (col.Metadata.Schema.GetColumnOrNull(MetadataUtils.Kinds.ScoreValueKind)?.Type == TextType.Instance)
                 {
-                    var res = ColumnInfo.CreateFromIndex(schema, col);
-                    if (testType(res.Type))
-                        return res;
+                    col.Metadata.GetValue(MetadataUtils.Kinds.ScoreValueKind, ref tmp);
+                    if (ReadOnlyMemoryUtils.EqualsStr(valueKind, tmp) && testType(col.Type))
+                        return col;
                 }
             }
 

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

@@ -134,9 +134,9 @@ Dictionary<string, IDataView> IEvaluator.Evaluate(RoleMappedData data)
         private protected virtual IEnumerable<KeyValuePair<RoleMappedSchema.ColumnRole, string>> GetInputColumnRolesCore(RoleMappedSchema schema)
         {
             // Get the score column information.
-            var scoreInfo = EvaluateUtils.GetScoreColumnInfo(Host, schema.Schema, ScoreCol, nameof(ArgumentsBase.ScoreColumn),
+            var scoreCol = EvaluateUtils.GetScoreColumn(Host, schema.Schema, ScoreCol, nameof(ArgumentsBase.ScoreColumn),
                 ScoreColumnKind);
-            yield return RoleMappedSchema.CreatePair(MetadataUtils.Const.ScoreValueKind.Score, scoreInfo.Name);
+            yield return RoleMappedSchema.CreatePair(MetadataUtils.Const.ScoreValueKind.Score, scoreCol.Name);
 
             // Get the label column information.
             string label = EvaluateUtils.GetColName(LabelCol, schema.Label, DefaultColumnNames.Label);

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

@@ -645,9 +645,9 @@ private protected override IEnumerable<string> GetPerInstanceColumnsToSave(RoleM
             {
                 yield return schema.Label.Value.Name;
 
-                var scoreInfo = EvaluateUtils.GetScoreColumnInfo(Host, schema.Schema, ScoreCol, nameof(Arguments.ScoreColumn),
+                var scoreCol = EvaluateUtils.GetScoreColumn(Host, schema.Schema, ScoreCol, nameof(Arguments.ScoreColumn),
                     MetadataUtils.Const.ScoreColumnKind.MultiOutputRegression);
-                yield return scoreInfo.Name;
+                yield return scoreCol.Name;
             }
 
             // Return the output columns.

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

@@ -545,9 +545,9 @@ private protected override IEnumerable<string> GetPerInstanceColumnsToSave(RoleM
 
             // The quantile regression evaluator outputs the label and score columns.
             yield return schema.Label.Value.Name;
-            var scoreInfo = EvaluateUtils.GetScoreColumnInfo(Host, schema.Schema, ScoreCol, nameof(Arguments.ScoreColumn),
+            var scoreCol = EvaluateUtils.GetScoreColumn(Host, schema.Schema, ScoreCol, nameof(Arguments.ScoreColumn),
                 MetadataUtils.Const.ScoreColumnKind.QuantileRegression);
-            yield return scoreInfo.Name;
+            yield return scoreCol.Name;
 
             // Return the output columns.
             yield return RegressionPerInstanceEvaluator.L1;

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

@@ -938,9 +938,9 @@ private protected override IEnumerable<string> GetPerInstanceColumnsToSave(RoleM
             // The ranking evaluator outputs the label, group key and score columns.
             yield return schema.Group.Value.Name;
             yield return schema.Label.Value.Name;
-            var scoreInfo = EvaluateUtils.GetScoreColumnInfo(Host, schema.Schema, ScoreCol, nameof(Arguments.ScoreColumn),
+            var scoreCol = EvaluateUtils.GetScoreColumn(Host, schema.Schema, ScoreCol, nameof(Arguments.ScoreColumn),
                 MetadataUtils.Const.ScoreColumnKind.Ranking);
-            yield return scoreInfo.Name;
+            yield return scoreCol.Name;
 
             // Return the output columns.
             yield return RankerPerInstanceTransform.Ndcg;

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

@@ -356,9 +356,9 @@ private protected override IEnumerable<string> GetPerInstanceColumnsToSave(RoleM
 
             // The regression evaluator outputs the label and score columns.
             yield return schema.Label.Value.Name;
-            var scoreInfo = EvaluateUtils.GetScoreColumnInfo(Host, schema.Schema, ScoreCol, nameof(Arguments.ScoreColumn),
+            var scoreCol = EvaluateUtils.GetScoreColumn(Host, schema.Schema, ScoreCol, nameof(Arguments.ScoreColumn),
                 MetadataUtils.Const.ScoreColumnKind.Regression);
-            yield return scoreInfo.Name;
+            yield return scoreCol.Name;
 
             // Return the output columns.
             yield return RegressionPerInstanceEvaluator.L1;

src/Microsoft.ML.Ensemble/Selector/SubModelSelector/BaseSubModelSelector.cs

@@ -107,26 +107,26 @@ public virtual void CalculateMetrics(FeatureSubsetModel<IPredictorProducing<TOut
             {
                 case PredictionKind.BinaryClassification:
                     yield return RoleMappedSchema.CreatePair(RoleMappedSchema.ColumnRole.Label, testSchema.Label.Value.Name);
-                    var scoreInfo = EvaluateUtils.GetScoreColumnInfo(Host, scoredSchema, null, nameof(BinaryClassifierMamlEvaluator.ArgumentsBase.ScoreColumn),
+                    var scoreCol = EvaluateUtils.GetScoreColumn(Host, scoredSchema, null, nameof(BinaryClassifierMamlEvaluator.ArgumentsBase.ScoreColumn),
                         MetadataUtils.Const.ScoreColumnKind.BinaryClassification);
-                    yield return RoleMappedSchema.CreatePair(MetadataUtils.Const.ScoreValueKind.Score, scoreInfo.Name);
+                    yield return RoleMappedSchema.CreatePair(MetadataUtils.Const.ScoreValueKind.Score, scoreCol.Name);
                     // Get the optional probability column.
-                    var probInfo = EvaluateUtils.GetOptAuxScoreColumnInfo(Host, scoredSchema, null, nameof(BinaryClassifierMamlEvaluator.Arguments.ProbabilityColumn),
-                        scoreInfo.Index, MetadataUtils.Const.ScoreValueKind.Probability, t => t == NumberType.Float);
-                    if (probInfo != null)
-                        yield return RoleMappedSchema.CreatePair(MetadataUtils.Const.ScoreValueKind.Probability, probInfo.Name);
+                    var probCol = EvaluateUtils.GetOptAuxScoreColumn(Host, scoredSchema, null, nameof(BinaryClassifierMamlEvaluator.Arguments.ProbabilityColumn),
+                        scoreCol.Index, MetadataUtils.Const.ScoreValueKind.Probability, NumberType.Float.Equals);
+                    if (probCol.HasValue)
+                        yield return RoleMappedSchema.CreatePair(MetadataUtils.Const.ScoreValueKind.Probability, probCol.Value.Name);
                     yield break;
                 case PredictionKind.Regression:
                     yield return RoleMappedSchema.CreatePair(RoleMappedSchema.ColumnRole.Label, testSchema.Label.Value.Name);
-                    scoreInfo = EvaluateUtils.GetScoreColumnInfo(Host, scoredSchema, null, nameof(RegressionMamlEvaluator.Arguments.ScoreColumn),
+                    scoreCol = EvaluateUtils.GetScoreColumn(Host, scoredSchema, null, nameof(RegressionMamlEvaluator.Arguments.ScoreColumn),
                         MetadataUtils.Const.ScoreColumnKind.Regression);
-                    yield return RoleMappedSchema.CreatePair(MetadataUtils.Const.ScoreValueKind.Score, scoreInfo.Name);
+                    yield return RoleMappedSchema.CreatePair(MetadataUtils.Const.ScoreValueKind.Score, scoreCol.Name);
                     yield break;
                 case PredictionKind.MultiClassClassification:
                     yield return RoleMappedSchema.CreatePair(RoleMappedSchema.ColumnRole.Label, testSchema.Label.Value.Name);
-                    scoreInfo = EvaluateUtils.GetScoreColumnInfo(Host, scoredSchema, null, nameof(MultiClassMamlEvaluator.Arguments.ScoreColumn),
+                    scoreCol = EvaluateUtils.GetScoreColumn(Host, scoredSchema, null, nameof(MultiClassMamlEvaluator.Arguments.ScoreColumn),
                         MetadataUtils.Const.ScoreColumnKind.MultiClassClassification);
-                    yield return RoleMappedSchema.CreatePair(MetadataUtils.Const.ScoreValueKind.Score, scoreInfo.Name);
+                    yield return RoleMappedSchema.CreatePair(MetadataUtils.Const.ScoreValueKind.Score, scoreCol.Name);
                     yield break;
                 default:
                     throw Host.Except("Unrecognized prediction kind '{0}'", PredictionKind);