PermutationFeatureImportanceExtensions.xml

<Type Name="PermutationFeatureImportanceExtensions" FullName="Microsoft.ML.PermutationFeatureImportanceExtensions">
  <TypeSignature Language="C#" Value="public static class PermutationFeatureImportanceExtensions" />
  <TypeSignature Language="ILAsm" Value=".class public auto ansi abstract sealed beforefieldinit PermutationFeatureImportanceExtensions extends System.Object" />
  <TypeSignature Language="DocId" Value="T:Microsoft.ML.PermutationFeatureImportanceExtensions" />
  <TypeSignature Language="VB.NET" Value="Public Module PermutationFeatureImportanceExtensions" />
  <TypeSignature Language="F#" Value="type PermutationFeatureImportanceExtensions = class" />
  <AssemblyInfo>
    <AssemblyName>Microsoft.ML.Transforms</AssemblyName>
    <AssemblyVersion>1.0.0.0</AssemblyVersion>
  </AssemblyInfo>
  <Base>
    <BaseTypeName>System.Object</BaseTypeName>
  </Base>
  <Interfaces />
  <Docs>
    <summary>
            Collection of extension methods used by <see cref="T:Microsoft.ML.RegressionCatalog" />,
             <see cref="T:Microsoft.ML.BinaryClassificationCatalog" />, <see cref="T:Microsoft.ML.MulticlassClassificationCatalog" />,
             and <see cref="T:Microsoft.ML.RankingCatalog" /> to create instances of permutation feature importance components.
            </summary>
    <remarks>To be added.</remarks>
  </Docs>
  <Members>
    <Member MemberName="PermutationFeatureImportance">
      <MemberSignature Language="C#" Value="public static System.Collections.Immutable.ImmutableDictionary&lt;string,Microsoft.ML.Data.MulticlassClassificationMetricsStatistics&gt; PermutationFeatureImportance (this Microsoft.ML.MulticlassClassificationCatalog catalog, Microsoft.ML.ITransformer model, Microsoft.ML.IDataView data, string labelColumnName = &quot;Label&quot;, bool useFeatureWeightFilter = false, int? numberOfExamplesToUse = default, int permutationCount = 1);" />
      <MemberSignature Language="ILAsm" Value=".method public static hidebysig class System.Collections.Immutable.ImmutableDictionary`2&lt;string, class Microsoft.ML.Data.MulticlassClassificationMetricsStatistics&gt; PermutationFeatureImportance(class Microsoft.ML.MulticlassClassificationCatalog catalog, class Microsoft.ML.ITransformer model, class Microsoft.ML.IDataView data, string labelColumnName, bool useFeatureWeightFilter, valuetype System.Nullable`1&lt;int32&gt; numberOfExamplesToUse, int32 permutationCount) cil managed" />
      <MemberSignature Language="DocId" Value="M:Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance(Microsoft.ML.MulticlassClassificationCatalog,Microsoft.ML.ITransformer,Microsoft.ML.IDataView,System.String,System.Boolean,System.Nullable{System.Int32},System.Int32)" />
      <MemberSignature Language="VB.NET" Value="&lt;Extension()&gt;&#xA;Public Function PermutationFeatureImportance (catalog As MulticlassClassificationCatalog, model As ITransformer, data As IDataView, Optional labelColumnName As String = &quot;Label&quot;, Optional useFeatureWeightFilter As Boolean = false, Optional numberOfExamplesToUse As Nullable(Of Integer) = Nothing, Optional permutationCount As Integer = 1) As ImmutableDictionary(Of String, MulticlassClassificationMetricsStatistics)" />
      <MemberSignature Language="F#" Value="static member PermutationFeatureImportance : Microsoft.ML.MulticlassClassificationCatalog * Microsoft.ML.ITransformer * Microsoft.ML.IDataView * string * bool * Nullable&lt;int&gt; * int -&gt; System.Collections.Immutable.ImmutableDictionary&lt;string, Microsoft.ML.Data.MulticlassClassificationMetricsStatistics&gt;" Usage="Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance (catalog, model, data, labelColumnName, useFeatureWeightFilter, numberOfExamplesToUse, permutationCount)" />
      <MemberType>Method</MemberType>
      <AssemblyInfo>
        <AssemblyName>Microsoft.ML.Transforms</AssemblyName>
        <AssemblyVersion>1.0.0.0</AssemblyVersion>
      </AssemblyInfo>
      <ReturnValue>
        <ReturnType>System.Collections.Immutable.ImmutableDictionary&lt;System.String,Microsoft.ML.Data.MulticlassClassificationMetricsStatistics&gt;</ReturnType>
      </ReturnValue>
      <Parameters>
        <Parameter Name="catalog" Type="Microsoft.ML.MulticlassClassificationCatalog" RefType="this" Index="0" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="model" Type="Microsoft.ML.ITransformer" Index="1" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="data" Type="Microsoft.ML.IDataView" Index="2" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="labelColumnName" Type="System.String" Index="3" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="useFeatureWeightFilter" Type="System.Boolean" Index="4" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="numberOfExamplesToUse" Type="System.Nullable&lt;System.Int32&gt;" Index="5" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="permutationCount" Type="System.Int32" Index="6" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
      </Parameters>
      <Docs>
        <param name="catalog">The multiclass classification catalog.</param>
        <param name="model">The model on which to evaluate feature importance.</param>
        <param name="data">The evaluation data set.</param>
        <param name="labelColumnName">Label column name. The column data must be <see cref="T:Microsoft.ML.Data.KeyDataViewType" />.</param>
        <param name="useFeatureWeightFilter">Use features weight to pre-filter features.</param>
        <param name="numberOfExamplesToUse">Limit the number of examples to evaluate on. <cref langword="null" /> means up to ~2 bln examples from <paramref param="data" /> will be used.</param>
        <param name="permutationCount">The number of permutations to perform.</param>
        <summary>
            Permutation Feature Importance (PFI) for MulticlassClassification.
            </summary>
        <returns>Dictionary mapping each feature to its per-feature 'contributions' to the score.</returns>
        <remarks>
          <para>
            Permutation feature importance (PFI) is a technique to determine the global importance of features in a trained
            machine learning model. PFI is a simple yet powerful technique motivated by Breiman in his Random Forest paper, section 10
            (Breiman. <a href="https://www.stat.berkeley.edu/~breiman/randomforest2001.pdf">"Random Forests."</a> Machine Learning, 2001.)
            The advantage of the PFI method is that it is model agnostic -- it works with any model that can be
            evaluated -- and it can use any dataset, not just the training set, to compute feature importance metrics.
            </para>
          <para>
            PFI works by taking a labeled dataset, choosing a feature, and permuting the values
            for that feature across all the examples, so that each example now has a random value for the feature and
            the original values for all other features. The evaluation metric (e.g. micro-accuracy) is then calculated
            for this modified dataset, and the change in the evaluation metric from the original dataset is computed.
            The larger the change in the evaluation metric, the more important the feature is to the model.
            PFI works by performing this permutation analysis across all the features of a model, one after another.
            </para>
          <para>
            In this implementation, PFI computes the change in all possible multiclass classification evaluation metrics for each feature, and an
            <see cref="T:System.Collections.Immutable.ImmutableArray" /> of <see cref="T:Microsoft.ML.Data.MulticlassClassificationMetrics" /> objects is returned. See the sample below for an
            example of working with these results to analyze the feature importance of a model.
            </para>
        </remarks>
        <example>
          <format type="text/markdown"><![CDATA[
            [!code-csharp[PermutationFeatureImportance](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/MulticlassClassification/PermutationFeatureImportance.cs)]
            ]]></format>
        </example>
      </Docs>
    </Member>
    <Member MemberName="PermutationFeatureImportance">
      <MemberSignature Language="C#" Value="public static System.Collections.Immutable.ImmutableDictionary&lt;string,Microsoft.ML.Data.RegressionMetricsStatistics&gt; PermutationFeatureImportance (this Microsoft.ML.RegressionCatalog catalog, Microsoft.ML.ITransformer model, Microsoft.ML.IDataView data, string labelColumnName = &quot;Label&quot;, bool useFeatureWeightFilter = false, int? numberOfExamplesToUse = default, int permutationCount = 1);" />
      <MemberSignature Language="ILAsm" Value=".method public static hidebysig class System.Collections.Immutable.ImmutableDictionary`2&lt;string, class Microsoft.ML.Data.RegressionMetricsStatistics&gt; PermutationFeatureImportance(class Microsoft.ML.RegressionCatalog catalog, class Microsoft.ML.ITransformer model, class Microsoft.ML.IDataView data, string labelColumnName, bool useFeatureWeightFilter, valuetype System.Nullable`1&lt;int32&gt; numberOfExamplesToUse, int32 permutationCount) cil managed" />
      <MemberSignature Language="DocId" Value="M:Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance(Microsoft.ML.RegressionCatalog,Microsoft.ML.ITransformer,Microsoft.ML.IDataView,System.String,System.Boolean,System.Nullable{System.Int32},System.Int32)" />
      <MemberSignature Language="VB.NET" Value="&lt;Extension()&gt;&#xA;Public Function PermutationFeatureImportance (catalog As RegressionCatalog, model As ITransformer, data As IDataView, Optional labelColumnName As String = &quot;Label&quot;, Optional useFeatureWeightFilter As Boolean = false, Optional numberOfExamplesToUse As Nullable(Of Integer) = Nothing, Optional permutationCount As Integer = 1) As ImmutableDictionary(Of String, RegressionMetricsStatistics)" />
      <MemberSignature Language="F#" Value="static member PermutationFeatureImportance : Microsoft.ML.RegressionCatalog * Microsoft.ML.ITransformer * Microsoft.ML.IDataView * string * bool * Nullable&lt;int&gt; * int -&gt; System.Collections.Immutable.ImmutableDictionary&lt;string, Microsoft.ML.Data.RegressionMetricsStatistics&gt;" Usage="Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance (catalog, model, data, labelColumnName, useFeatureWeightFilter, numberOfExamplesToUse, permutationCount)" />
      <MemberType>Method</MemberType>
      <AssemblyInfo>
        <AssemblyName>Microsoft.ML.Transforms</AssemblyName>
        <AssemblyVersion>1.0.0.0</AssemblyVersion>
      </AssemblyInfo>
      <ReturnValue>
        <ReturnType>System.Collections.Immutable.ImmutableDictionary&lt;System.String,Microsoft.ML.Data.RegressionMetricsStatistics&gt;</ReturnType>
      </ReturnValue>
      <Parameters>
        <Parameter Name="catalog" Type="Microsoft.ML.RegressionCatalog" RefType="this" Index="0" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="model" Type="Microsoft.ML.ITransformer" Index="1" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="data" Type="Microsoft.ML.IDataView" Index="2" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="labelColumnName" Type="System.String" Index="3" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="useFeatureWeightFilter" Type="System.Boolean" Index="4" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="numberOfExamplesToUse" Type="System.Nullable&lt;System.Int32&gt;" Index="5" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="permutationCount" Type="System.Int32" Index="6" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
      </Parameters>
      <Docs>
        <param name="catalog">The regression catalog.</param>
        <param name="model">The model on which to evaluate feature importance.</param>
        <param name="data">The evaluation data set.</param>
        <param name="labelColumnName">Label column name. The column data must be <see cref="T:System.Single" />.</param>
        <param name="useFeatureWeightFilter">Use features weight to pre-filter features.</param>
        <param name="numberOfExamplesToUse">Limit the number of examples to evaluate on. <cref langword="null" /> means up to ~2 bln examples from <paramref param="data" /> will be used.</param>
        <param name="permutationCount">The number of permutations to perform.</param>
        <summary>
            Permutation Feature Importance (PFI) for Regression.
            </summary>
        <returns>Dictionary mapping each feature to its per-feature 'contributions' to the score.</returns>
        <remarks>
          <para>
            Permutation feature importance (PFI) is a technique to determine the global importance of features in a trained
            machine learning model. PFI is a simple yet powerful technique motivated by Breiman in his Random Forest paper, section 10
            (Breiman. <a href="https://www.stat.berkeley.edu/~breiman/randomforest2001.pdf">"Random Forests."</a> Machine Learning, 2001.)
            The advantage of the PFI method is that it is model agnostic -- it works with any model that can be
            evaluated -- and it can use any dataset, not just the training set, to compute feature importance metrics.
            </para>
          <para>
            PFI works by taking a labeled dataset, choosing a feature, and permuting the values
            for that feature across all the examples, so that each example now has a random value for the feature and
            the original values for all other features. The evaluation metric (e.g. R-squared) is then calculated
            for this modified dataset, and the change in the evaluation metric from the original dataset is computed.
            The larger the change in the evaluation metric, the more important the feature is to the model.
            PFI works by performing this permutation analysis across all the features of a model, one after another.
            </para>
          <para>
            In this implementation, PFI computes the change in all possible regression evaluation metrics for each feature, and an
            <see cref="T:System.Collections.Immutable.ImmutableArray" /> of <see cref="T:Microsoft.ML.Data.RegressionMetrics" /> objects is returned. See the sample below for an
            example of working with these results to analyze the feature importance of a model.
            </para>
        </remarks>
        <example>
          <format type="text/markdown"><![CDATA[
            [!code-csharp[PermutationFeatureImportance](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/Regression/PermutationFeatureImportance.cs)]
            ]]></format>
        </example>
      </Docs>
    </Member>
    <Member MemberName="PermutationFeatureImportance">
      <MemberSignature Language="C#" Value="public static System.Collections.Immutable.ImmutableDictionary&lt;string,Microsoft.ML.Data.RankingMetricsStatistics&gt; PermutationFeatureImportance (this Microsoft.ML.RankingCatalog catalog, Microsoft.ML.ITransformer model, Microsoft.ML.IDataView data, string labelColumnName = &quot;Label&quot;, string rowGroupColumnName = &quot;GroupId&quot;, bool useFeatureWeightFilter = false, int? numberOfExamplesToUse = default, int permutationCount = 1);" />
      <MemberSignature Language="ILAsm" Value=".method public static hidebysig class System.Collections.Immutable.ImmutableDictionary`2&lt;string, class Microsoft.ML.Data.RankingMetricsStatistics&gt; PermutationFeatureImportance(class Microsoft.ML.RankingCatalog catalog, class Microsoft.ML.ITransformer model, class Microsoft.ML.IDataView data, string labelColumnName, string rowGroupColumnName, bool useFeatureWeightFilter, valuetype System.Nullable`1&lt;int32&gt; numberOfExamplesToUse, int32 permutationCount) cil managed" />
      <MemberSignature Language="DocId" Value="M:Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance(Microsoft.ML.RankingCatalog,Microsoft.ML.ITransformer,Microsoft.ML.IDataView,System.String,System.String,System.Boolean,System.Nullable{System.Int32},System.Int32)" />
      <MemberSignature Language="VB.NET" Value="&lt;Extension()&gt;&#xA;Public Function PermutationFeatureImportance (catalog As RankingCatalog, model As ITransformer, data As IDataView, Optional labelColumnName As String = &quot;Label&quot;, Optional rowGroupColumnName As String = &quot;GroupId&quot;, Optional useFeatureWeightFilter As Boolean = false, Optional numberOfExamplesToUse As Nullable(Of Integer) = Nothing, Optional permutationCount As Integer = 1) As ImmutableDictionary(Of String, RankingMetricsStatistics)" />
      <MemberSignature Language="F#" Value="static member PermutationFeatureImportance : Microsoft.ML.RankingCatalog * Microsoft.ML.ITransformer * Microsoft.ML.IDataView * string * string * bool * Nullable&lt;int&gt; * int -&gt; System.Collections.Immutable.ImmutableDictionary&lt;string, Microsoft.ML.Data.RankingMetricsStatistics&gt;" Usage="Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance (catalog, model, data, labelColumnName, rowGroupColumnName, useFeatureWeightFilter, numberOfExamplesToUse, permutationCount)" />
      <MemberType>Method</MemberType>
      <AssemblyInfo>
        <AssemblyName>Microsoft.ML.Transforms</AssemblyName>
        <AssemblyVersion>1.0.0.0</AssemblyVersion>
      </AssemblyInfo>
      <ReturnValue>
        <ReturnType>System.Collections.Immutable.ImmutableDictionary&lt;System.String,Microsoft.ML.Data.RankingMetricsStatistics&gt;</ReturnType>
      </ReturnValue>
      <Parameters>
        <Parameter Name="catalog" Type="Microsoft.ML.RankingCatalog" RefType="this" Index="0" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="model" Type="Microsoft.ML.ITransformer" Index="1" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="data" Type="Microsoft.ML.IDataView" Index="2" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="labelColumnName" Type="System.String" Index="3" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="rowGroupColumnName" Type="System.String" Index="4" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="useFeatureWeightFilter" Type="System.Boolean" Index="5" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="numberOfExamplesToUse" Type="System.Nullable&lt;System.Int32&gt;" Index="6" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="permutationCount" Type="System.Int32" Index="7" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
      </Parameters>
      <Docs>
        <param name="catalog">The ranking catalog.</param>
        <param name="model">The model on which to evaluate feature importance.</param>
        <param name="data">The evaluation data set.</param>
        <param name="labelColumnName">Label column name. The column data must be <see cref="T:System.Single" /> or <see cref="T:Microsoft.ML.Data.KeyDataViewType" />.</param>
        <param name="rowGroupColumnName">GroupId column name</param>
        <param name="useFeatureWeightFilter">Use features weight to pre-filter features.</param>
        <param name="numberOfExamplesToUse">Limit the number of examples to evaluate on. <cref langword="null" /> means up to ~2 bln examples from <paramref param="data" /> will be used.</param>
        <param name="permutationCount">The number of permutations to perform.</param>
        <summary>
            Permutation Feature Importance (PFI) for Ranking.
            </summary>
        <returns>Dictionary mapping each feature to its per-feature 'contributions' to the score.</returns>
        <remarks>
          <para>
            Permutation feature importance (PFI) is a technique to determine the global importance of features in a trained
            machine learning model. PFI is a simple yet powerful technique motivated by Breiman in his Random Forest paper, section 10
            (Breiman. <a href="https://www.stat.berkeley.edu/~breiman/randomforest2001.pdf">"Random Forests."</a> Machine Learning, 2001.)
            The advantage of the PFI method is that it is model agnostic -- it works with any model that can be
            evaluated -- and it can use any dataset, not just the training set, to compute feature importance metrics.
            </para>
          <para>
            PFI works by taking a labeled dataset, choosing a feature, and permuting the values
            for that feature across all the examples, so that each example now has a random value for the feature and
            the original values for all other features. The evaluation metric (e.g. NDCG) is then calculated
            for this modified dataset, and the change in the evaluation metric from the original dataset is computed.
            The larger the change in the evaluation metric, the more important the feature is to the model.
            PFI works by performing this permutation analysis across all the features of a model, one after another.
            </para>
          <para>
            In this implementation, PFI computes the change in all possible ranking evaluation metrics for each feature, and an
            <see cref="T:System.Collections.Immutable.ImmutableArray" /> of <see cref="T:Microsoft.ML.Data.RankingMetrics" /> objects is returned. See the sample below for an
            example of working with these results to analyze the feature importance of a model.
            </para>
        </remarks>
        <example>
          <format type="text/markdown"><![CDATA[
            [!code-csharp[PermutationFeatureImportance](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/Ranking/PermutationFeatureImportance.cs)]
            ]]></format>
        </example>
      </Docs>
    </Member>
    <Member MemberName="PermutationFeatureImportance&lt;TModel&gt;">
      <MemberSignature Language="C#" Value="public static System.Collections.Immutable.ImmutableArray&lt;Microsoft.ML.Data.BinaryClassificationMetricsStatistics&gt; PermutationFeatureImportance&lt;TModel&gt; (this Microsoft.ML.BinaryClassificationCatalog catalog, Microsoft.ML.ISingleFeaturePredictionTransformer&lt;TModel&gt; predictionTransformer, Microsoft.ML.IDataView data, string labelColumnName = &quot;Label&quot;, bool useFeatureWeightFilter = false, int? numberOfExamplesToUse = default, int permutationCount = 1) where TModel : class;" />
      <MemberSignature Language="ILAsm" Value=".method public static hidebysig valuetype System.Collections.Immutable.ImmutableArray`1&lt;class Microsoft.ML.Data.BinaryClassificationMetricsStatistics&gt; PermutationFeatureImportance&lt;class TModel&gt;(class Microsoft.ML.BinaryClassificationCatalog catalog, class Microsoft.ML.ISingleFeaturePredictionTransformer`1&lt;!!TModel&gt; predictionTransformer, class Microsoft.ML.IDataView data, string labelColumnName, bool useFeatureWeightFilter, valuetype System.Nullable`1&lt;int32&gt; numberOfExamplesToUse, int32 permutationCount) cil managed" />
      <MemberSignature Language="DocId" Value="M:Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance``1(Microsoft.ML.BinaryClassificationCatalog,Microsoft.ML.ISingleFeaturePredictionTransformer{``0},Microsoft.ML.IDataView,System.String,System.Boolean,System.Nullable{System.Int32},System.Int32)" />
      <MemberSignature Language="VB.NET" Value="&lt;Extension()&gt;&#xA;Public Function PermutationFeatureImportance(Of TModel As Class) (catalog As BinaryClassificationCatalog, predictionTransformer As ISingleFeaturePredictionTransformer(Of TModel), data As IDataView, Optional labelColumnName As String = &quot;Label&quot;, Optional useFeatureWeightFilter As Boolean = false, Optional numberOfExamplesToUse As Nullable(Of Integer) = Nothing, Optional permutationCount As Integer = 1) As ImmutableArray(Of BinaryClassificationMetricsStatistics)" />
      <MemberSignature Language="F#" Value="static member PermutationFeatureImportance : Microsoft.ML.BinaryClassificationCatalog * Microsoft.ML.ISingleFeaturePredictionTransformer&lt;'Model (requires 'Model : null)&gt; * Microsoft.ML.IDataView * string * bool * Nullable&lt;int&gt; * int -&gt; System.Collections.Immutable.ImmutableArray&lt;Microsoft.ML.Data.BinaryClassificationMetricsStatistics&gt; (requires 'Model : null)" Usage="Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance (catalog, predictionTransformer, data, labelColumnName, useFeatureWeightFilter, numberOfExamplesToUse, permutationCount)" />
      <MemberType>Method</MemberType>
      <AssemblyInfo>
        <AssemblyName>Microsoft.ML.Transforms</AssemblyName>
        <AssemblyVersion>1.0.0.0</AssemblyVersion>
      </AssemblyInfo>
      <ReturnValue>
        <ReturnType>System.Collections.Immutable.ImmutableArray&lt;Microsoft.ML.Data.BinaryClassificationMetricsStatistics&gt;</ReturnType>
      </ReturnValue>
      <TypeParameters>
        <TypeParameter Name="TModel">
          <Constraints>
            <ParameterAttribute>ReferenceTypeConstraint</ParameterAttribute>
          </Constraints>
        </TypeParameter>
      </TypeParameters>
      <Parameters>
        <Parameter Name="catalog" Type="Microsoft.ML.BinaryClassificationCatalog" RefType="this" />
        <Parameter Name="predictionTransformer" Type="Microsoft.ML.ISingleFeaturePredictionTransformer&lt;TModel&gt;" />
        <Parameter Name="data" Type="Microsoft.ML.IDataView" />
        <Parameter Name="labelColumnName" Type="System.String" />
        <Parameter Name="useFeatureWeightFilter" Type="System.Boolean" />
        <Parameter Name="numberOfExamplesToUse" Type="System.Nullable&lt;System.Int32&gt;" />
        <Parameter Name="permutationCount" Type="System.Int32" />
      </Parameters>
      <Docs>
        <typeparam name="TModel">To be added.</typeparam>
        <param name="catalog">The binary classification catalog.</param>
        <param name="predictionTransformer">The model on which to evaluate feature importance.</param>
        <param name="data">The evaluation data set.</param>
        <param name="labelColumnName">Label column name. The column data must be <see cref="T:System.Boolean" />.</param>
        <param name="useFeatureWeightFilter">Use features weight to pre-filter features.</param>
        <param name="numberOfExamplesToUse">Limit the number of examples to evaluate on. <cref langword="null" /> means up to ~2 bln examples from <paramref param="data" /> will be used.</param>
        <param name="permutationCount">The number of permutations to perform.</param>
        <summary>
            Permutation Feature Importance (PFI) for Binary Classification.
            </summary>
        <returns>Array of per-feature 'contributions' to the score.</returns>
        <remarks>
          <para>
            Permutation feature importance (PFI) is a technique to determine the global importance of features in a trained
            machine learning model. PFI is a simple yet powerful technique motivated by Breiman in his Random Forest paper, section 10
            (Breiman. <a href="https://www.stat.berkeley.edu/~breiman/randomforest2001.pdf">"Random Forests."</a> Machine Learning, 2001.)
            The advantage of the PFI method is that it is model agnostic -- it works with any model that can be
            evaluated -- and it can use any dataset, not just the training set, to compute feature importance metrics.
            </para>
          <para>
            PFI works by taking a labeled dataset, choosing a feature, and permuting the values
            for that feature across all the examples, so that each example now has a random value for the feature and
            the original values for all other features. The evaluation metric (e.g. AUC) is then calculated
            for this modified dataset, and the change in the evaluation metric from the original dataset is computed.
            The larger the change in the evaluation metric, the more important the feature is to the model.
            PFI works by performing this permutation analysis across all the features of a model, one after another.
            </para>
          <para>
            In this implementation, PFI computes the change in all possible binary classification evaluation metrics for each feature, and an
            <see cref="T:System.Collections.Immutable.ImmutableArray" /> of <see cref="T:Microsoft.ML.Data.BinaryClassificationMetrics" /> objects is returned. See the sample below for an
            example of working with these results to analyze the feature importance of a model.
            </para>
        </remarks>
        <example>
          <format type="text/markdown"><![CDATA[
            [!code-csharp[PermutationFeatureImportance](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/BinaryClassification/PermutationFeatureImportance.cs)]
            ]]></format>
        </example>
      </Docs>
    </Member>
    <Member MemberName="PermutationFeatureImportance&lt;TModel&gt;">
      <MemberSignature Language="C#" Value="public static System.Collections.Immutable.ImmutableArray&lt;Microsoft.ML.Data.MulticlassClassificationMetricsStatistics&gt; PermutationFeatureImportance&lt;TModel&gt; (this Microsoft.ML.MulticlassClassificationCatalog catalog, Microsoft.ML.ISingleFeaturePredictionTransformer&lt;TModel&gt; predictionTransformer, Microsoft.ML.IDataView data, string labelColumnName = &quot;Label&quot;, bool useFeatureWeightFilter = false, int? numberOfExamplesToUse = default, int permutationCount = 1) where TModel : class;" />
      <MemberSignature Language="ILAsm" Value=".method public static hidebysig valuetype System.Collections.Immutable.ImmutableArray`1&lt;class Microsoft.ML.Data.MulticlassClassificationMetricsStatistics&gt; PermutationFeatureImportance&lt;class TModel&gt;(class Microsoft.ML.MulticlassClassificationCatalog catalog, class Microsoft.ML.ISingleFeaturePredictionTransformer`1&lt;!!TModel&gt; predictionTransformer, class Microsoft.ML.IDataView data, string labelColumnName, bool useFeatureWeightFilter, valuetype System.Nullable`1&lt;int32&gt; numberOfExamplesToUse, int32 permutationCount) cil managed" />
      <MemberSignature Language="DocId" Value="M:Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance``1(Microsoft.ML.MulticlassClassificationCatalog,Microsoft.ML.ISingleFeaturePredictionTransformer{``0},Microsoft.ML.IDataView,System.String,System.Boolean,System.Nullable{System.Int32},System.Int32)" />
      <MemberSignature Language="VB.NET" Value="&lt;Extension()&gt;&#xA;Public Function PermutationFeatureImportance(Of TModel As Class) (catalog As MulticlassClassificationCatalog, predictionTransformer As ISingleFeaturePredictionTransformer(Of TModel), data As IDataView, Optional labelColumnName As String = &quot;Label&quot;, Optional useFeatureWeightFilter As Boolean = false, Optional numberOfExamplesToUse As Nullable(Of Integer) = Nothing, Optional permutationCount As Integer = 1) As ImmutableArray(Of MulticlassClassificationMetricsStatistics)" />
      <MemberSignature Language="F#" Value="static member PermutationFeatureImportance : Microsoft.ML.MulticlassClassificationCatalog * Microsoft.ML.ISingleFeaturePredictionTransformer&lt;'Model (requires 'Model : null)&gt; * Microsoft.ML.IDataView * string * bool * Nullable&lt;int&gt; * int -&gt; System.Collections.Immutable.ImmutableArray&lt;Microsoft.ML.Data.MulticlassClassificationMetricsStatistics&gt; (requires 'Model : null)" Usage="Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance (catalog, predictionTransformer, data, labelColumnName, useFeatureWeightFilter, numberOfExamplesToUse, permutationCount)" />
      <MemberType>Method</MemberType>
      <AssemblyInfo>
        <AssemblyName>Microsoft.ML.Transforms</AssemblyName>
        <AssemblyVersion>1.0.0.0</AssemblyVersion>
      </AssemblyInfo>
      <ReturnValue>
        <ReturnType>System.Collections.Immutable.ImmutableArray&lt;Microsoft.ML.Data.MulticlassClassificationMetricsStatistics&gt;</ReturnType>
      </ReturnValue>
      <TypeParameters>
        <TypeParameter Name="TModel">
          <Constraints>
            <ParameterAttribute>ReferenceTypeConstraint</ParameterAttribute>
          </Constraints>
        </TypeParameter>
      </TypeParameters>
      <Parameters>
        <Parameter Name="catalog" Type="Microsoft.ML.MulticlassClassificationCatalog" RefType="this" />
        <Parameter Name="predictionTransformer" Type="Microsoft.ML.ISingleFeaturePredictionTransformer&lt;TModel&gt;" />
        <Parameter Name="data" Type="Microsoft.ML.IDataView" />
        <Parameter Name="labelColumnName" Type="System.String" />
        <Parameter Name="useFeatureWeightFilter" Type="System.Boolean" />
        <Parameter Name="numberOfExamplesToUse" Type="System.Nullable&lt;System.Int32&gt;" />
        <Parameter Name="permutationCount" Type="System.Int32" />
      </Parameters>
      <Docs>
        <typeparam name="TModel">To be added.</typeparam>
        <param name="catalog">The multiclass classification catalog.</param>
        <param name="predictionTransformer">The model on which to evaluate feature importance.</param>
        <param name="data">The evaluation data set.</param>
        <param name="labelColumnName">Label column name. The column data must be <see cref="T:Microsoft.ML.Data.KeyDataViewType" />.</param>
        <param name="useFeatureWeightFilter">Use features weight to pre-filter features.</param>
        <param name="numberOfExamplesToUse">Limit the number of examples to evaluate on. <cref langword="null" /> means up to ~2 bln examples from <paramref param="data" /> will be used.</param>
        <param name="permutationCount">The number of permutations to perform.</param>
        <summary>
            Permutation Feature Importance (PFI) for MulticlassClassification.
            </summary>
        <returns>Array of per-feature 'contributions' to the score.</returns>
        <remarks>
          <para>
            Permutation feature importance (PFI) is a technique to determine the global importance of features in a trained
            machine learning model. PFI is a simple yet powerful technique motivated by Breiman in his Random Forest paper, section 10
            (Breiman. <a href="https://www.stat.berkeley.edu/~breiman/randomforest2001.pdf">"Random Forests."</a> Machine Learning, 2001.)
            The advantage of the PFI method is that it is model agnostic -- it works with any model that can be
            evaluated -- and it can use any dataset, not just the training set, to compute feature importance metrics.
            </para>
          <para>
            PFI works by taking a labeled dataset, choosing a feature, and permuting the values
            for that feature across all the examples, so that each example now has a random value for the feature and
            the original values for all other features. The evaluation metric (e.g. micro-accuracy) is then calculated
            for this modified dataset, and the change in the evaluation metric from the original dataset is computed.
            The larger the change in the evaluation metric, the more important the feature is to the model.
            PFI works by performing this permutation analysis across all the features of a model, one after another.
            </para>
          <para>
            In this implementation, PFI computes the change in all possible multiclass classification evaluation metrics for each feature, and an
            <see cref="T:System.Collections.Immutable.ImmutableArray" /> of <see cref="T:Microsoft.ML.Data.MulticlassClassificationMetrics" /> objects is returned. See the sample below for an
            example of working with these results to analyze the feature importance of a model.
            </para>
        </remarks>
        <example>
          <format type="text/markdown"><![CDATA[
            [!code-csharp[PermutationFeatureImportance](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/MulticlassClassification/PermutationFeatureImportance.cs)]
            ]]></format>
        </example>
      </Docs>
    </Member>
    <Member MemberName="PermutationFeatureImportance&lt;TModel&gt;">
      <MemberSignature Language="C#" Value="public static System.Collections.Immutable.ImmutableArray&lt;Microsoft.ML.Data.RegressionMetricsStatistics&gt; PermutationFeatureImportance&lt;TModel&gt; (this Microsoft.ML.RegressionCatalog catalog, Microsoft.ML.ISingleFeaturePredictionTransformer&lt;TModel&gt; predictionTransformer, Microsoft.ML.IDataView data, string labelColumnName = &quot;Label&quot;, bool useFeatureWeightFilter = false, int? numberOfExamplesToUse = default, int permutationCount = 1) where TModel : class;" />
      <MemberSignature Language="ILAsm" Value=".method public static hidebysig valuetype System.Collections.Immutable.ImmutableArray`1&lt;class Microsoft.ML.Data.RegressionMetricsStatistics&gt; PermutationFeatureImportance&lt;class TModel&gt;(class Microsoft.ML.RegressionCatalog catalog, class Microsoft.ML.ISingleFeaturePredictionTransformer`1&lt;!!TModel&gt; predictionTransformer, class Microsoft.ML.IDataView data, string labelColumnName, bool useFeatureWeightFilter, valuetype System.Nullable`1&lt;int32&gt; numberOfExamplesToUse, int32 permutationCount) cil managed" />
      <MemberSignature Language="DocId" Value="M:Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance``1(Microsoft.ML.RegressionCatalog,Microsoft.ML.ISingleFeaturePredictionTransformer{``0},Microsoft.ML.IDataView,System.String,System.Boolean,System.Nullable{System.Int32},System.Int32)" />
      <MemberSignature Language="VB.NET" Value="&lt;Extension()&gt;&#xA;Public Function PermutationFeatureImportance(Of TModel As Class) (catalog As RegressionCatalog, predictionTransformer As ISingleFeaturePredictionTransformer(Of TModel), data As IDataView, Optional labelColumnName As String = &quot;Label&quot;, Optional useFeatureWeightFilter As Boolean = false, Optional numberOfExamplesToUse As Nullable(Of Integer) = Nothing, Optional permutationCount As Integer = 1) As ImmutableArray(Of RegressionMetricsStatistics)" />
      <MemberSignature Language="F#" Value="static member PermutationFeatureImportance : Microsoft.ML.RegressionCatalog * Microsoft.ML.ISingleFeaturePredictionTransformer&lt;'Model (requires 'Model : null)&gt; * Microsoft.ML.IDataView * string * bool * Nullable&lt;int&gt; * int -&gt; System.Collections.Immutable.ImmutableArray&lt;Microsoft.ML.Data.RegressionMetricsStatistics&gt; (requires 'Model : null)" Usage="Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance (catalog, predictionTransformer, data, labelColumnName, useFeatureWeightFilter, numberOfExamplesToUse, permutationCount)" />
      <MemberType>Method</MemberType>
      <AssemblyInfo>
        <AssemblyName>Microsoft.ML.Transforms</AssemblyName>
        <AssemblyVersion>1.0.0.0</AssemblyVersion>
      </AssemblyInfo>
      <ReturnValue>
        <ReturnType>System.Collections.Immutable.ImmutableArray&lt;Microsoft.ML.Data.RegressionMetricsStatistics&gt;</ReturnType>
      </ReturnValue>
      <TypeParameters>
        <TypeParameter Name="TModel">
          <Constraints>
            <ParameterAttribute>ReferenceTypeConstraint</ParameterAttribute>
          </Constraints>
        </TypeParameter>
      </TypeParameters>
      <Parameters>
        <Parameter Name="catalog" Type="Microsoft.ML.RegressionCatalog" RefType="this" />
        <Parameter Name="predictionTransformer" Type="Microsoft.ML.ISingleFeaturePredictionTransformer&lt;TModel&gt;" />
        <Parameter Name="data" Type="Microsoft.ML.IDataView" />
        <Parameter Name="labelColumnName" Type="System.String" />
        <Parameter Name="useFeatureWeightFilter" Type="System.Boolean" />
        <Parameter Name="numberOfExamplesToUse" Type="System.Nullable&lt;System.Int32&gt;" />
        <Parameter Name="permutationCount" Type="System.Int32" />
      </Parameters>
      <Docs>
        <typeparam name="TModel">To be added.</typeparam>
        <param name="catalog">The regression catalog.</param>
        <param name="predictionTransformer">The model on which to evaluate feature importance.</param>
        <param name="data">The evaluation data set.</param>
        <param name="labelColumnName">Label column name. The column data must be <see cref="T:System.Single" />.</param>
        <param name="useFeatureWeightFilter">Use features weight to pre-filter features.</param>
        <param name="numberOfExamplesToUse">Limit the number of examples to evaluate on. <cref langword="null" /> means up to ~2 bln examples from <paramref param="data" /> will be used.</param>
        <param name="permutationCount">The number of permutations to perform.</param>
        <summary>
            Permutation Feature Importance (PFI) for Regression.
            </summary>
        <returns>Array of per-feature 'contributions' to the score.</returns>
        <remarks>
          <para>
            Permutation feature importance (PFI) is a technique to determine the global importance of features in a trained
            machine learning model. PFI is a simple yet powerful technique motivated by Breiman in his Random Forest paper, section 10
            (Breiman. <a href="https://www.stat.berkeley.edu/~breiman/randomforest2001.pdf">"Random Forests."</a> Machine Learning, 2001.)
            The advantage of the PFI method is that it is model agnostic -- it works with any model that can be
            evaluated -- and it can use any dataset, not just the training set, to compute feature importance metrics.
            </para>
          <para>
            PFI works by taking a labeled dataset, choosing a feature, and permuting the values
            for that feature across all the examples, so that each example now has a random value for the feature and
            the original values for all other features. The evaluation metric (e.g. R-squared) is then calculated
            for this modified dataset, and the change in the evaluation metric from the original dataset is computed.
            The larger the change in the evaluation metric, the more important the feature is to the model.
            PFI works by performing this permutation analysis across all the features of a model, one after another.
            </para>
          <para>
            In this implementation, PFI computes the change in all possible regression evaluation metrics for each feature, and an
            <see cref="T:System.Collections.Immutable.ImmutableArray" /> of <see cref="T:Microsoft.ML.Data.RegressionMetrics" /> objects is returned. See the sample below for an
            example of working with these results to analyze the feature importance of a model.
            </para>
        </remarks>
        <example>
          <format type="text/markdown"><![CDATA[
            [!code-csharp[PermutationFeatureImportance](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/Regression/PermutationFeatureImportance.cs)]
            ]]></format>
        </example>
      </Docs>
    </Member>
    <Member MemberName="PermutationFeatureImportance&lt;TModel&gt;">
      <MemberSignature Language="C#" Value="public static System.Collections.Immutable.ImmutableArray&lt;Microsoft.ML.Data.RankingMetricsStatistics&gt; PermutationFeatureImportance&lt;TModel&gt; (this Microsoft.ML.RankingCatalog catalog, Microsoft.ML.ISingleFeaturePredictionTransformer&lt;TModel&gt; predictionTransformer, Microsoft.ML.IDataView data, string labelColumnName = &quot;Label&quot;, string rowGroupColumnName = &quot;GroupId&quot;, bool useFeatureWeightFilter = false, int? numberOfExamplesToUse = default, int permutationCount = 1) where TModel : class;" />
      <MemberSignature Language="ILAsm" Value=".method public static hidebysig valuetype System.Collections.Immutable.ImmutableArray`1&lt;class Microsoft.ML.Data.RankingMetricsStatistics&gt; PermutationFeatureImportance&lt;class TModel&gt;(class Microsoft.ML.RankingCatalog catalog, class Microsoft.ML.ISingleFeaturePredictionTransformer`1&lt;!!TModel&gt; predictionTransformer, class Microsoft.ML.IDataView data, string labelColumnName, string rowGroupColumnName, bool useFeatureWeightFilter, valuetype System.Nullable`1&lt;int32&gt; numberOfExamplesToUse, int32 permutationCount) cil managed" />
      <MemberSignature Language="DocId" Value="M:Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance``1(Microsoft.ML.RankingCatalog,Microsoft.ML.ISingleFeaturePredictionTransformer{``0},Microsoft.ML.IDataView,System.String,System.String,System.Boolean,System.Nullable{System.Int32},System.Int32)" />
      <MemberSignature Language="VB.NET" Value="&lt;Extension()&gt;&#xA;Public Function PermutationFeatureImportance(Of TModel As Class) (catalog As RankingCatalog, predictionTransformer As ISingleFeaturePredictionTransformer(Of TModel), data As IDataView, Optional labelColumnName As String = &quot;Label&quot;, Optional rowGroupColumnName As String = &quot;GroupId&quot;, Optional useFeatureWeightFilter As Boolean = false, Optional numberOfExamplesToUse As Nullable(Of Integer) = Nothing, Optional permutationCount As Integer = 1) As ImmutableArray(Of RankingMetricsStatistics)" />
      <MemberSignature Language="F#" Value="static member PermutationFeatureImportance : Microsoft.ML.RankingCatalog * Microsoft.ML.ISingleFeaturePredictionTransformer&lt;'Model (requires 'Model : null)&gt; * Microsoft.ML.IDataView * string * string * bool * Nullable&lt;int&gt; * int -&gt; System.Collections.Immutable.ImmutableArray&lt;Microsoft.ML.Data.RankingMetricsStatistics&gt; (requires 'Model : null)" Usage="Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance (catalog, predictionTransformer, data, labelColumnName, rowGroupColumnName, useFeatureWeightFilter, numberOfExamplesToUse, permutationCount)" />
      <MemberType>Method</MemberType>
      <AssemblyInfo>
        <AssemblyName>Microsoft.ML.Transforms</AssemblyName>
        <AssemblyVersion>1.0.0.0</AssemblyVersion>
      </AssemblyInfo>
      <ReturnValue>
        <ReturnType>System.Collections.Immutable.ImmutableArray&lt;Microsoft.ML.Data.RankingMetricsStatistics&gt;</ReturnType>
      </ReturnValue>
      <TypeParameters>
        <TypeParameter Name="TModel">
          <Constraints>
            <ParameterAttribute>ReferenceTypeConstraint</ParameterAttribute>
          </Constraints>
        </TypeParameter>
      </TypeParameters>
      <Parameters>
        <Parameter Name="catalog" Type="Microsoft.ML.RankingCatalog" RefType="this" />
        <Parameter Name="predictionTransformer" Type="Microsoft.ML.ISingleFeaturePredictionTransformer&lt;TModel&gt;" />
        <Parameter Name="data" Type="Microsoft.ML.IDataView" />
        <Parameter Name="labelColumnName" Type="System.String" />
        <Parameter Name="rowGroupColumnName" Type="System.String" />
        <Parameter Name="useFeatureWeightFilter" Type="System.Boolean" />
        <Parameter Name="numberOfExamplesToUse" Type="System.Nullable&lt;System.Int32&gt;" />
        <Parameter Name="permutationCount" Type="System.Int32" />
      </Parameters>
      <Docs>
        <typeparam name="TModel">To be added.</typeparam>
        <param name="catalog">The ranking catalog.</param>
        <param name="predictionTransformer">The model on which to evaluate feature importance.</param>
        <param name="data">The evaluation data set.</param>
        <param name="labelColumnName">Label column name. The column data must be <see cref="T:System.Single" /> or <see cref="T:Microsoft.ML.Data.KeyDataViewType" />.</param>
        <param name="rowGroupColumnName">GroupId column name</param>
        <param name="useFeatureWeightFilter">Use features weight to pre-filter features.</param>
        <param name="numberOfExamplesToUse">Limit the number of examples to evaluate on. <cref langword="null" /> means up to ~2 bln examples from <paramref param="data" /> will be used.</param>
        <param name="permutationCount">The number of permutations to perform.</param>
        <summary>
            Permutation Feature Importance (PFI) for Ranking.
            </summary>
        <returns>Array of per-feature 'contributions' to the score.</returns>
        <remarks>
          <para>
            Permutation feature importance (PFI) is a technique to determine the global importance of features in a trained
            machine learning model. PFI is a simple yet powerful technique motivated by Breiman in his Random Forest paper, section 10
            (Breiman. <a href="https://www.stat.berkeley.edu/~breiman/randomforest2001.pdf">"Random Forests."</a> Machine Learning, 2001.)
            The advantage of the PFI method is that it is model agnostic -- it works with any model that can be
            evaluated -- and it can use any dataset, not just the training set, to compute feature importance metrics.
            </para>
          <para>
            PFI works by taking a labeled dataset, choosing a feature, and permuting the values
            for that feature across all the examples, so that each example now has a random value for the feature and
            the original values for all other features. The evaluation metric (e.g. NDCG) is then calculated
            for this modified dataset, and the change in the evaluation metric from the original dataset is computed.
            The larger the change in the evaluation metric, the more important the feature is to the model.
            PFI works by performing this permutation analysis across all the features of a model, one after another.
            </para>
          <para>
            In this implementation, PFI computes the change in all possible ranking evaluation metrics for each feature, and an
            <see cref="T:System.Collections.Immutable.ImmutableArray" /> of <see cref="T:Microsoft.ML.Data.RankingMetrics" /> objects is returned. See the sample below for an
            example of working with these results to analyze the feature importance of a model.
            </para>
        </remarks>
        <example>
          <format type="text/markdown"><![CDATA[
            [!code-csharp[PermutationFeatureImportance](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/Ranking/PermutationFeatureImportance.cs)]
            ]]></format>
        </example>
      </Docs>
    </Member>
    <Member MemberName="PermutationFeatureImportanceNonCalibrated">
      <MemberSignature Language="C#" Value="public static System.Collections.Immutable.ImmutableDictionary&lt;string,Microsoft.ML.Data.BinaryClassificationMetricsStatistics&gt; PermutationFeatureImportanceNonCalibrated (this Microsoft.ML.BinaryClassificationCatalog catalog, Microsoft.ML.ITransformer model, Microsoft.ML.IDataView data, string labelColumnName = &quot;Label&quot;, bool useFeatureWeightFilter = false, int? numberOfExamplesToUse = default, int permutationCount = 1);" />
      <MemberSignature Language="ILAsm" Value=".method public static hidebysig class System.Collections.Immutable.ImmutableDictionary`2&lt;string, class Microsoft.ML.Data.BinaryClassificationMetricsStatistics&gt; PermutationFeatureImportanceNonCalibrated(class Microsoft.ML.BinaryClassificationCatalog catalog, class Microsoft.ML.ITransformer model, class Microsoft.ML.IDataView data, string labelColumnName, bool useFeatureWeightFilter, valuetype System.Nullable`1&lt;int32&gt; numberOfExamplesToUse, int32 permutationCount) cil managed" />
      <MemberSignature Language="DocId" Value="M:Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportanceNonCalibrated(Microsoft.ML.BinaryClassificationCatalog,Microsoft.ML.ITransformer,Microsoft.ML.IDataView,System.String,System.Boolean,System.Nullable{System.Int32},System.Int32)" />
      <MemberSignature Language="VB.NET" Value="&lt;Extension()&gt;&#xA;Public Function PermutationFeatureImportanceNonCalibrated (catalog As BinaryClassificationCatalog, model As ITransformer, data As IDataView, Optional labelColumnName As String = &quot;Label&quot;, Optional useFeatureWeightFilter As Boolean = false, Optional numberOfExamplesToUse As Nullable(Of Integer) = Nothing, Optional permutationCount As Integer = 1) As ImmutableDictionary(Of String, BinaryClassificationMetricsStatistics)" />
      <MemberSignature Language="F#" Value="static member PermutationFeatureImportanceNonCalibrated : Microsoft.ML.BinaryClassificationCatalog * Microsoft.ML.ITransformer * Microsoft.ML.IDataView * string * bool * Nullable&lt;int&gt; * int -&gt; System.Collections.Immutable.ImmutableDictionary&lt;string, Microsoft.ML.Data.BinaryClassificationMetricsStatistics&gt;" Usage="Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportanceNonCalibrated (catalog, model, data, labelColumnName, useFeatureWeightFilter, numberOfExamplesToUse, permutationCount)" />
      <MemberType>Method</MemberType>
      <AssemblyInfo>
        <AssemblyName>Microsoft.ML.Transforms</AssemblyName>
        <AssemblyVersion>1.0.0.0</AssemblyVersion>
      </AssemblyInfo>
      <ReturnValue>
        <ReturnType>System.Collections.Immutable.ImmutableDictionary&lt;System.String,Microsoft.ML.Data.BinaryClassificationMetricsStatistics&gt;</ReturnType>
      </ReturnValue>
      <Parameters>
        <Parameter Name="catalog" Type="Microsoft.ML.BinaryClassificationCatalog" RefType="this" Index="0" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="model" Type="Microsoft.ML.ITransformer" Index="1" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="data" Type="Microsoft.ML.IDataView" Index="2" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="labelColumnName" Type="System.String" Index="3" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="useFeatureWeightFilter" Type="System.Boolean" Index="4" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="numberOfExamplesToUse" Type="System.Nullable&lt;System.Int32&gt;" Index="5" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
        <Parameter Name="permutationCount" Type="System.Int32" Index="6" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
      </Parameters>
      <Docs>
        <param name="catalog">The binary classification catalog.</param>
        <param name="model">The model on which to evaluate feature importance.</param>
        <param name="data">The evaluation data set.</param>
        <param name="labelColumnName">Label column name. The column data must be <see cref="T:System.Boolean" />.</param>
        <param name="useFeatureWeightFilter">Use features weight to pre-filter features.</param>
        <param name="numberOfExamplesToUse">Limit the number of examples to evaluate on. <cref langword="null" /> means up to ~2 bln examples from <paramref param="data" /> will be used.</param>
        <param name="permutationCount">The number of permutations to perform.</param>
        <summary>
            Permutation Feature Importance (PFI) for Binary Classification.
            </summary>
        <returns>Dictionary mapping each feature to its per-feature 'contributions' to the score.</returns>
        <remarks>
          <para>
            Permutation feature importance (PFI) is a technique to determine the global importance of features in a trained
            machine learning model. PFI is a simple yet powerful technique motivated by Breiman in his Random Forest paper, section 10
            (Breiman. <a href="https://www.stat.berkeley.edu/~breiman/randomforest2001.pdf">"Random Forests."</a> Machine Learning, 2001.)
            The advantage of the PFI method is that it is model agnostic -- it works with any model that can be
            evaluated -- and it can use any dataset, not just the training set, to compute feature importance metrics.
            </para>
          <para>
            PFI works by taking a labeled dataset, choosing a feature, and permuting the values
            for that feature across all the examples, so that each example now has a random value for the feature and
            the original values for all other features. The evaluation metric (e.g. AUC) is then calculated
            for this modified dataset, and the change in the evaluation metric from the original dataset is computed.
            The larger the change in the evaluation metric, the more important the feature is to the model.
            PFI works by performing this permutation analysis across all the features of a model, one after another.
            </para>
          <para>
            In this implementation, PFI computes the change in all possible binary classification evaluation metrics for each feature, and an
            <see cref="T:System.Collections.Immutable.ImmutableArray" /> of <see cref="T:Microsoft.ML.Data.BinaryClassificationMetrics" /> objects is returned. See the sample below for an
            example of working with these results to analyze the feature importance of a model.
            </para>
        </remarks>
        <example>
          <format type="text/markdown"><![CDATA[
            [!code-csharp[PermutationFeatureImportance](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/BinaryClassification/PermutationFeatureImportance.cs)]
            ]]></format>
        </example>
      </Docs>
    </Member>
  </Members>
</Type>