dev/modules/compose.html



<!DOCTYPE html>
<!-- data-theme below is forced to be "light" but should be changed if we use pydata-theme-sphinx in the future -->
<!--[if IE 8]><html class="no-js lt-ie9" lang="en" data-content_root="../" data-theme="light"> <![endif]-->
<!--[if gt IE 8]><!--> <html class="no-js" lang="en" data-content_root="../" data-theme="light"> <!--<![endif]-->
<head>
  <meta charset="utf-8">
  <meta name="viewport" content="width=device-width, initial-scale=1" />
<meta property="og:title" content="6.1. Pipelines and composite estimators" />
<meta property="og:type" content="website" />
<meta property="og:url" content="https://scikit-learn/stable/modules/compose.html" />
<meta property="og:site_name" content="scikit-learn" />
<meta property="og:description" content="To build a composite estimator, transformers are usually combined with other transformers or with predictors(such as classifiers or regressors). The most common tool used for composing estimators i..." />
<meta property="og:image" content="https://scikit-learn.org/stable/_static/scikit-learn-logo-small.png" />
<meta property="og:image:alt" content="scikit-learn" />
<meta name="description" content="To build a composite estimator, transformers are usually combined with other transformers or with predictors(such as classifiers or regressors). The most common tool used for composing estimators i..." />

  <meta name="viewport" content="width=device-width, initial-scale=1.0">

  
  <title>6.1. Pipelines and composite estimators &mdash; scikit-learn 1.5.dev0 documentation</title>
  
  <link rel="canonical" href="https://scikit-learn.org/stable/modules/compose.html" />

  
  <link rel="shortcut icon" href="../_static/favicon.ico"/>
  

  <link rel="stylesheet" href="../_static/css/vendor/bootstrap.min.css" type="text/css" />
  <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
  <link rel="stylesheet" href="../_static/css/theme.css" type="text/css" />
  <link rel="stylesheet" href="../_static/copybutton.css" type="text/css" />
  <link rel="stylesheet" href="../_static/plot_directive.css" type="text/css" />
  <link rel="stylesheet" href="../https://fonts.googleapis.com/css?family=Vibur" type="text/css" />
  <link rel="stylesheet" href="../_static/jupyterlite_sphinx.css" type="text/css" />
  <link rel="stylesheet" href="../_static/sg_gallery.css" type="text/css" />
  <link rel="stylesheet" href="../_static/sg_gallery-binder.css" type="text/css" />
  <link rel="stylesheet" href="../_static/sg_gallery-dataframe.css" type="text/css" />
  <link rel="stylesheet" href="../_static/sg_gallery-rendered-html.css" type="text/css" />
  <link rel="stylesheet" href="../_static/css/theme.css" type="text/css" />
<script id="documentation_options" data-url_root="../" src="../_static/documentation_options.js"></script>
<script src="../_static/js/vendor/jquery-3.6.3.slim.min.js"></script>
<script src="../_static/js/details-permalink.js"></script> 
</head>
<body>


<nav id="navbar" class="sk-docs-navbar navbar navbar-expand-md navbar-light bg-light py-0">
  <div class="container-fluid sk-docs-container px-0">
      <a class="navbar-brand py-0" href="../index.html">
        <img
          class="sk-brand-img"
          src="../_static/scikit-learn-logo-small.png"
          alt="logo"/>
      </a>
    <button
      id="sk-navbar-toggler"
      class="navbar-toggler"
      type="button"
      data-toggle="collapse"
      data-target="#navbarSupportedContent"
      aria-controls="navbarSupportedContent"
      aria-expanded="false"
      aria-label="Toggle navigation"
    >
      <span class="navbar-toggler-icon"></span>
    </button>

    <div class="sk-navbar-collapse collapse navbar-collapse" id="navbarSupportedContent">
      <ul class="navbar-nav mr-auto">
        <li class="nav-item">
          <a class="sk-nav-link nav-link" href="../install.html">Install</a>
        </li>
        <li class="nav-item">
          <a class="sk-nav-link nav-link" href="../user_guide.html">User Guide</a>
        </li>
        <li class="nav-item">
          <a class="sk-nav-link nav-link" href="classes.html">API</a>
        </li>
        <li class="nav-item">
          <a class="sk-nav-link nav-link" href="../auto_examples/index.html">Examples</a>
        </li>
        <li class="nav-item">
          <a class="sk-nav-link nav-link" target="_blank" rel="noopener noreferrer" href="https://blog.scikit-learn.org/">Community</a>
        </li>
        <li class="nav-item">
          <a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../getting_started.html" >Getting Started</a>
        </li>
        <li class="nav-item">
          <a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../tutorial/index.html" >Tutorial</a>
        </li>
        <li class="nav-item">
          <a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../whats_new/v1.5.html" >What's new</a>
        </li>
        <li class="nav-item">
          <a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../glossary.html" >Glossary</a>
        </li>
        <li class="nav-item">
          <a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../developers/index.html" >Development</a>
        </li>
        <li class="nav-item">
          <a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../faq.html" >FAQ</a>
        </li>
        <li class="nav-item">
          <a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../support.html" >Support</a>
        </li>
        <li class="nav-item">
          <a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../related_projects.html" >Related packages</a>
        </li>
        <li class="nav-item">
          <a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../roadmap.html" >Roadmap</a>
        </li>
        <li class="nav-item">
          <a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../governance.html" >Governance</a>
        </li>
        <li class="nav-item">
          <a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../about.html" >About us</a>
        </li>
        <li class="nav-item">
          <a class="sk-nav-link nav-link nav-more-item-mobile-items" href="https://github.com/scikit-learn/scikit-learn" >GitHub</a>
        </li>
        <li class="nav-item">
          <a class="sk-nav-link nav-link nav-more-item-mobile-items" href="https://scikit-learn.org/dev/versions.html" >Other Versions and Download</a>
        </li>
        <li class="nav-item dropdown nav-more-item-dropdown">
          <a class="sk-nav-link nav-link dropdown-toggle" href="#" id="navbarDropdown" role="button" data-toggle="dropdown" aria-haspopup="true" aria-expanded="false">More</a>
          <div class="dropdown-menu" aria-labelledby="navbarDropdown">
              <a class="sk-nav-dropdown-item dropdown-item" href="../getting_started.html" >Getting Started</a>
              <a class="sk-nav-dropdown-item dropdown-item" href="../tutorial/index.html" >Tutorial</a>
              <a class="sk-nav-dropdown-item dropdown-item" href="../whats_new/v1.5.html" >What's new</a>
              <a class="sk-nav-dropdown-item dropdown-item" href="../glossary.html" >Glossary</a>
              <a class="sk-nav-dropdown-item dropdown-item" href="../developers/index.html" >Development</a>
              <a class="sk-nav-dropdown-item dropdown-item" href="../faq.html" >FAQ</a>
              <a class="sk-nav-dropdown-item dropdown-item" href="../support.html" >Support</a>
              <a class="sk-nav-dropdown-item dropdown-item" href="../related_projects.html" >Related packages</a>
              <a class="sk-nav-dropdown-item dropdown-item" href="../roadmap.html" >Roadmap</a>
              <a class="sk-nav-dropdown-item dropdown-item" href="../governance.html" >Governance</a>
              <a class="sk-nav-dropdown-item dropdown-item" href="../about.html" >About us</a>
              <a class="sk-nav-dropdown-item dropdown-item" href="https://github.com/scikit-learn/scikit-learn" >GitHub</a>
              <a class="sk-nav-dropdown-item dropdown-item" href="https://scikit-learn.org/dev/versions.html" >Other Versions and Download</a>
          </div>
        </li>
      </ul>
      <div id="searchbox" role="search">
          <div class="searchformwrapper">
          <form class="search" action="../search.html" method="get">
            <input class="sk-search-text-input" type="text" name="q" aria-labelledby="searchlabel" />
            <input class="sk-search-text-btn" type="submit" value="Go" />
          </form>
          </div>
      </div>
    </div>
  </div>
</nav>
<div class="d-flex" id="sk-doc-wrapper">
    <input type="checkbox" name="sk-toggle-checkbox" id="sk-toggle-checkbox">
    <label id="sk-sidemenu-toggle" class="sk-btn-toggle-toc btn sk-btn-primary" for="sk-toggle-checkbox">Toggle Menu</label>
    <div id="sk-sidebar-wrapper" class="border-right">
      <div class="sk-sidebar-toc-wrapper">
        <div class="btn-group w-100 mb-2" role="group" aria-label="rellinks">
            <a href="../data_transforms.html" role="button" class="btn sk-btn-rellink py-1" sk-rellink-tooltip="6. Dataset transformations">Prev</a><a href="../data_transforms.html" role="button" class="btn sk-btn-rellink py-1" sk-rellink-tooltip="6. Dataset transformations">Up</a>
            <a href="feature_extraction.html" role="button" class="btn sk-btn-rellink py-1" sk-rellink-tooltip="6.2. Feature extraction">Next</a>
        </div>
        <div class="alert alert-danger p-1 mb-2" role="alert">
          <p class="text-center mb-0">
          <strong>scikit-learn 1.5.dev0</strong><br/>
          <a href="https://scikit-learn.org/dev/versions.html">Other versions</a>
          </p>
        </div>
        <div class="alert alert-warning p-1 mb-2" role="alert">
          <p class="text-center mb-0">
            Please <a class="font-weight-bold" href="../about.html#citing-scikit-learn"><string>cite us</string></a> if you use the software.
          </p>
        </div>
            <div class="sk-sidebar-toc">
              <ul>
<li><a class="reference internal" href="#">6.1. Pipelines and composite estimators</a><ul>
<li><a class="reference internal" href="#pipeline-chaining-estimators">6.1.1. Pipeline: chaining estimators</a><ul>
<li><a class="reference internal" href="#usage">6.1.1.1. Usage</a><ul>
<li><a class="reference internal" href="#build-a-pipeline">6.1.1.1.1. Build a pipeline</a></li>
<li><a class="reference internal" href="#access-pipeline-steps">6.1.1.1.2. Access pipeline steps</a></li>
<li><a class="reference internal" href="#tracking-feature-names-in-a-pipeline">6.1.1.1.3. Tracking feature names in a pipeline</a></li>
<li><a class="reference internal" href="#access-to-nested-parameters">6.1.1.1.4. Access to nested parameters</a></li>
</ul>
</li>
<li><a class="reference internal" href="#caching-transformers-avoid-repeated-computation">6.1.1.2. Caching transformers: avoid repeated computation</a></li>
</ul>
</li>
<li><a class="reference internal" href="#transforming-target-in-regression">6.1.2. Transforming target in regression</a></li>
<li><a class="reference internal" href="#featureunion-composite-feature-spaces">6.1.3. FeatureUnion: composite feature spaces</a><ul>
<li><a class="reference internal" href="#id1">6.1.3.1. Usage</a></li>
</ul>
</li>
<li><a class="reference internal" href="#columntransformer-for-heterogeneous-data">6.1.4. ColumnTransformer for heterogeneous data</a></li>
<li><a class="reference internal" href="#visualizing-composite-estimators">6.1.5. Visualizing Composite Estimators</a></li>
</ul>
</li>
</ul>

            </div>
      </div>
    </div>
    <div id="sk-page-content-wrapper">
      <div class="sk-page-content container-fluid body px-md-3" role="main">
        
  <section id="pipelines-and-composite-estimators">
<span id="combining-estimators"></span><h1><span class="section-number">6.1. </span>Pipelines and composite estimators<a class="headerlink" href="#pipelines-and-composite-estimators" title="Link to this heading">¶</a></h1>
<p>To build a composite estimator, transformers are usually combined with other
transformers or with <a class="reference internal" href="../glossary.html#term-predictors"><span class="xref std std-term">predictors</span></a> (such as classifiers or regressors).
The most common tool used for composing estimators is a <a class="reference internal" href="#pipeline"><span class="std std-ref">Pipeline</span></a>. Pipelines require all steps except the last to be a
<a class="reference internal" href="../glossary.html#term-transformer"><span class="xref std std-term">transformer</span></a>. The last step can be anything, a transformer, a
<a class="reference internal" href="../glossary.html#term-predictor"><span class="xref std std-term">predictor</span></a>, or a clustering estimator which might have or not have a
<code class="docutils literal notranslate"><span class="pre">.predict(...)</span></code> method. A pipeline exposes all methods provided by the last
estimator: if the last step provides a <code class="docutils literal notranslate"><span class="pre">transform</span></code> method, then the pipeline
would have a <code class="docutils literal notranslate"><span class="pre">transform</span></code> method and behave like a transformer. If the last step
provides a <code class="docutils literal notranslate"><span class="pre">predict</span></code> method, then the pipeline would expose that method, and
given a data <a class="reference internal" href="../glossary.html#term-X"><span class="xref std std-term">X</span></a>, use all steps except the last to transform the data,
and then give that transformed data to the <code class="docutils literal notranslate"><span class="pre">predict</span></code> method of the last step of
the pipeline. The class <code class="xref py py-class docutils literal notranslate"><span class="pre">Pipeline</span></code> is often used in combination with
<a class="reference internal" href="#column-transformer"><span class="std std-ref">ColumnTransformer</span></a> or
<a class="reference internal" href="#feature-union"><span class="std std-ref">FeatureUnion</span></a> which concatenate the output of transformers
into a composite feature space.
<a class="reference internal" href="#transformed-target-regressor"><span class="std std-ref">TransformedTargetRegressor</span></a>
deals with transforming the <a class="reference internal" href="../glossary.html#term-target"><span class="xref std std-term">target</span></a> (i.e. log-transform <a class="reference internal" href="../glossary.html#term-y"><span class="xref std std-term">y</span></a>).</p>
<section id="pipeline-chaining-estimators">
<span id="pipeline"></span><h2><span class="section-number">6.1.1. </span>Pipeline: chaining estimators<a class="headerlink" href="#pipeline-chaining-estimators" title="Link to this heading">¶</a></h2>
<p><a class="reference internal" href="generated/sklearn.pipeline.Pipeline.html#sklearn.pipeline.Pipeline" title="sklearn.pipeline.Pipeline"><code class="xref py py-class docutils literal notranslate"><span class="pre">Pipeline</span></code></a> can be used to chain multiple estimators
into one. This is useful as there is often a fixed sequence
of steps in processing the data, for example feature selection, normalization
and classification. <a class="reference internal" href="generated/sklearn.pipeline.Pipeline.html#sklearn.pipeline.Pipeline" title="sklearn.pipeline.Pipeline"><code class="xref py py-class docutils literal notranslate"><span class="pre">Pipeline</span></code></a> serves multiple purposes here:</p>
<dl class="simple">
<dt>Convenience and encapsulation</dt><dd><p>You only have to call <a class="reference internal" href="../glossary.html#term-fit"><span class="xref std std-term">fit</span></a> and <a class="reference internal" href="../glossary.html#term-predict"><span class="xref std std-term">predict</span></a> once on your
data to fit a whole sequence of estimators.</p>
</dd>
<dt>Joint parameter selection</dt><dd><p>You can <a class="reference internal" href="grid_search.html#grid-search"><span class="std std-ref">grid search</span></a>
over parameters of all estimators in the pipeline at once.</p>
</dd>
<dt>Safety</dt><dd><p>Pipelines help avoid leaking statistics from your test data into the
trained model in cross-validation, by ensuring that the same samples are
used to train the transformers and predictors.</p>
</dd>
</dl>
<p>All estimators in a pipeline, except the last one, must be transformers
(i.e. must have a <a class="reference internal" href="../glossary.html#term-transform"><span class="xref std std-term">transform</span></a> method).
The last estimator may be any type (transformer, classifier, etc.).</p>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>Calling <code class="docutils literal notranslate"><span class="pre">fit</span></code> on the pipeline is the same as calling <code class="docutils literal notranslate"><span class="pre">fit</span></code> on
each estimator in turn, <code class="docutils literal notranslate"><span class="pre">transform</span></code> the input and pass it on to the next step.
The pipeline has all the methods that the last estimator in the pipeline has,
i.e. if the last estimator is a classifier, the <a class="reference internal" href="generated/sklearn.pipeline.Pipeline.html#sklearn.pipeline.Pipeline" title="sklearn.pipeline.Pipeline"><code class="xref py py-class docutils literal notranslate"><span class="pre">Pipeline</span></code></a> can be used
as a classifier. If the last estimator is a transformer, again, so is the
pipeline.</p>
</div>
<section id="usage">
<h3><span class="section-number">6.1.1.1. </span>Usage<a class="headerlink" href="#usage" title="Link to this heading">¶</a></h3>
<section id="build-a-pipeline">
<h4><span class="section-number">6.1.1.1.1. </span>Build a pipeline<a class="headerlink" href="#build-a-pipeline" title="Link to this heading">¶</a></h4>
<p>The <a class="reference internal" href="generated/sklearn.pipeline.Pipeline.html#sklearn.pipeline.Pipeline" title="sklearn.pipeline.Pipeline"><code class="xref py py-class docutils literal notranslate"><span class="pre">Pipeline</span></code></a> is built using a list of <code class="docutils literal notranslate"><span class="pre">(key,</span> <span class="pre">value)</span></code> pairs, where
the <code class="docutils literal notranslate"><span class="pre">key</span></code> is a string containing the name you want to give this step and <code class="docutils literal notranslate"><span class="pre">value</span></code>
is an estimator object:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.pipeline</span> <span class="kn">import</span> <span class="n">Pipeline</span>
<span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.svm</span> <span class="kn">import</span> <span class="n">SVC</span>
<span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.decomposition</span> <span class="kn">import</span> <span class="n">PCA</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">estimators</span> <span class="o">=</span> <span class="p">[(</span><span class="s1">&#39;reduce_dim&#39;</span><span class="p">,</span> <span class="n">PCA</span><span class="p">()),</span> <span class="p">(</span><span class="s1">&#39;clf&#39;</span><span class="p">,</span> <span class="n">SVC</span><span class="p">())]</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">pipe</span> <span class="o">=</span> <span class="n">Pipeline</span><span class="p">(</span><span class="n">estimators</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">pipe</span>
<span class="go">Pipeline(steps=[(&#39;reduce_dim&#39;, PCA()), (&#39;clf&#39;, SVC())])</span>
</pre></div>
</div>
<p><details id="summary-anchor">
<summary class="btn btn-light">
<strong>Shorthand version using :func:`make_pipeline`</strong>
<span class="tooltiptext">Click for more details</span>
<a class="headerlink" href="#summary-anchor" title="Permalink to this heading">¶</a>
</summary>
<div class="card"></p>
<p>The utility function <a class="reference internal" href="generated/sklearn.pipeline.make_pipeline.html#sklearn.pipeline.make_pipeline" title="sklearn.pipeline.make_pipeline"><code class="xref py py-func docutils literal notranslate"><span class="pre">make_pipeline</span></code></a> is a shorthand
for constructing pipelines;
it takes a variable number of estimators and returns a pipeline,
filling in the names automatically:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.pipeline</span> <span class="kn">import</span> <span class="n">make_pipeline</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">make_pipeline</span><span class="p">(</span><span class="n">PCA</span><span class="p">(),</span> <span class="n">SVC</span><span class="p">())</span>
<span class="go">Pipeline(steps=[(&#39;pca&#39;, PCA()), (&#39;svc&#39;, SVC())])</span>
</pre></div>
</div>
<p></div>
</details></p>
</section>
<section id="access-pipeline-steps">
<h4><span class="section-number">6.1.1.1.2. </span>Access pipeline steps<a class="headerlink" href="#access-pipeline-steps" title="Link to this heading">¶</a></h4>
<p>The estimators of a pipeline are stored as a list in the <code class="docutils literal notranslate"><span class="pre">steps</span></code> attribute.
A sub-pipeline can be extracted using the slicing notation commonly used
for Python Sequences such as lists or strings (although only a step of 1 is
permitted). This is convenient for performing only some of the transformations
(or their inverse):</p>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">pipe</span><span class="p">[:</span><span class="mi">1</span><span class="p">]</span>
<span class="go">Pipeline(steps=[(&#39;reduce_dim&#39;, PCA())])</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">pipe</span><span class="p">[</span><span class="o">-</span><span class="mi">1</span><span class="p">:]</span>
<span class="go">Pipeline(steps=[(&#39;clf&#39;, SVC())])</span>
</pre></div>
</div>
<p><details id="summary-anchor">
<summary class="btn btn-light">
<strong>Accessing a step by name or position</strong>
<span class="tooltiptext">Click for more details</span>
<a class="headerlink" href="#summary-anchor" title="Permalink to this heading">¶</a>
</summary>
<div class="card"></p>
<p>A specific step can also be accessed by index or name by indexing (with <code class="docutils literal notranslate"><span class="pre">[idx]</span></code>) the
pipeline:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">pipe</span><span class="o">.</span><span class="n">steps</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span>
<span class="go">(&#39;reduce_dim&#39;, PCA())</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">pipe</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span>
<span class="go">PCA()</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">pipe</span><span class="p">[</span><span class="s1">&#39;reduce_dim&#39;</span><span class="p">]</span>
<span class="go">PCA()</span>
</pre></div>
</div>
<p><code class="docutils literal notranslate"><span class="pre">Pipeline</span></code>’s <code class="docutils literal notranslate"><span class="pre">named_steps</span></code> attribute allows accessing steps by name with tab
completion in interactive environments:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">pipe</span><span class="o">.</span><span class="n">named_steps</span><span class="o">.</span><span class="n">reduce_dim</span> <span class="ow">is</span> <span class="n">pipe</span><span class="p">[</span><span class="s1">&#39;reduce_dim&#39;</span><span class="p">]</span>
<span class="go">True</span>
</pre></div>
</div>
<p></div>
</details></p>
</section>
<section id="tracking-feature-names-in-a-pipeline">
<h4><span class="section-number">6.1.1.1.3. </span>Tracking feature names in a pipeline<a class="headerlink" href="#tracking-feature-names-in-a-pipeline" title="Link to this heading">¶</a></h4>
<p>To enable model inspection, <a class="reference internal" href="generated/sklearn.pipeline.Pipeline.html#sklearn.pipeline.Pipeline" title="sklearn.pipeline.Pipeline"><code class="xref py py-class docutils literal notranslate"><span class="pre">Pipeline</span></code></a> has a
<code class="docutils literal notranslate"><span class="pre">get_feature_names_out()</span></code> method, just like all transformers. You can use
pipeline slicing to get the feature names going into each step:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.datasets</span> <span class="kn">import</span> <span class="n">load_iris</span>
<span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.linear_model</span> <span class="kn">import</span> <span class="n">LogisticRegression</span>
<span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.feature_selection</span> <span class="kn">import</span> <span class="n">SelectKBest</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">iris</span> <span class="o">=</span> <span class="n">load_iris</span><span class="p">()</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">pipe</span> <span class="o">=</span> <span class="n">Pipeline</span><span class="p">(</span><span class="n">steps</span><span class="o">=</span><span class="p">[</span>
<span class="gp">... </span>   <span class="p">(</span><span class="s1">&#39;select&#39;</span><span class="p">,</span> <span class="n">SelectKBest</span><span class="p">(</span><span class="n">k</span><span class="o">=</span><span class="mi">2</span><span class="p">)),</span>
<span class="gp">... </span>   <span class="p">(</span><span class="s1">&#39;clf&#39;</span><span class="p">,</span> <span class="n">LogisticRegression</span><span class="p">())])</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">pipe</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">iris</span><span class="o">.</span><span class="n">data</span><span class="p">,</span> <span class="n">iris</span><span class="o">.</span><span class="n">target</span><span class="p">)</span>
<span class="go">Pipeline(steps=[(&#39;select&#39;, SelectKBest(...)), (&#39;clf&#39;, LogisticRegression(...))])</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">pipe</span><span class="p">[:</span><span class="o">-</span><span class="mi">1</span><span class="p">]</span><span class="o">.</span><span class="n">get_feature_names_out</span><span class="p">()</span>
<span class="go">array([&#39;x2&#39;, &#39;x3&#39;], ...)</span>
</pre></div>
</div>
<p><details id="summary-anchor">
<summary class="btn btn-light">
<strong>Customize feature names</strong>
<span class="tooltiptext">Click for more details</span>
<a class="headerlink" href="#summary-anchor" title="Permalink to this heading">¶</a>
</summary>
<div class="card"></p>
<p>You can also provide custom feature names for the input data using
<code class="docutils literal notranslate"><span class="pre">get_feature_names_out</span></code>:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">pipe</span><span class="p">[:</span><span class="o">-</span><span class="mi">1</span><span class="p">]</span><span class="o">.</span><span class="n">get_feature_names_out</span><span class="p">(</span><span class="n">iris</span><span class="o">.</span><span class="n">feature_names</span><span class="p">)</span>
<span class="go">array([&#39;petal length (cm)&#39;, &#39;petal width (cm)&#39;], ...)</span>
</pre></div>
</div>
<p></div>
</details></p>
</section>
<section id="access-to-nested-parameters">
<span id="pipeline-nested-parameters"></span><h4><span class="section-number">6.1.1.1.4. </span>Access to nested parameters<a class="headerlink" href="#access-to-nested-parameters" title="Link to this heading">¶</a></h4>
<p>It is common to adjust the parameters of an estimator within a pipeline. This parameter
is therefore nested because it belongs to a particular sub-step. Parameters of the
estimators in the pipeline are accessible using the <code class="docutils literal notranslate"><span class="pre">&lt;estimator&gt;__&lt;parameter&gt;</span></code>
syntax:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">pipe</span> <span class="o">=</span> <span class="n">Pipeline</span><span class="p">(</span><span class="n">steps</span><span class="o">=</span><span class="p">[(</span><span class="s2">&quot;reduce_dim&quot;</span><span class="p">,</span> <span class="n">PCA</span><span class="p">()),</span> <span class="p">(</span><span class="s2">&quot;clf&quot;</span><span class="p">,</span> <span class="n">SVC</span><span class="p">())])</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">pipe</span><span class="o">.</span><span class="n">set_params</span><span class="p">(</span><span class="n">clf__C</span><span class="o">=</span><span class="mi">10</span><span class="p">)</span>
<span class="go">Pipeline(steps=[(&#39;reduce_dim&#39;, PCA()), (&#39;clf&#39;, SVC(C=10))])</span>
</pre></div>
</div>
<p><details id="summary-anchor">
<summary class="btn btn-light">
<strong>When does it matter?</strong>
<span class="tooltiptext">Click for more details</span>
<a class="headerlink" href="#summary-anchor" title="Permalink to this heading">¶</a>
</summary>
<div class="card"></p>
<p>This is particularly important for doing grid searches:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.model_selection</span> <span class="kn">import</span> <span class="n">GridSearchCV</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">param_grid</span> <span class="o">=</span> <span class="nb">dict</span><span class="p">(</span><span class="n">reduce_dim__n_components</span><span class="o">=</span><span class="p">[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">5</span><span class="p">,</span> <span class="mi">10</span><span class="p">],</span>
<span class="gp">... </span>                  <span class="n">clf__C</span><span class="o">=</span><span class="p">[</span><span class="mf">0.1</span><span class="p">,</span> <span class="mi">10</span><span class="p">,</span> <span class="mi">100</span><span class="p">])</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">grid_search</span> <span class="o">=</span> <span class="n">GridSearchCV</span><span class="p">(</span><span class="n">pipe</span><span class="p">,</span> <span class="n">param_grid</span><span class="o">=</span><span class="n">param_grid</span><span class="p">)</span>
</pre></div>
</div>
<p>Individual steps may also be replaced as parameters, and non-final steps may be
ignored by setting them to <code class="docutils literal notranslate"><span class="pre">'passthrough'</span></code>:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">param_grid</span> <span class="o">=</span> <span class="nb">dict</span><span class="p">(</span><span class="n">reduce_dim</span><span class="o">=</span><span class="p">[</span><span class="s1">&#39;passthrough&#39;</span><span class="p">,</span> <span class="n">PCA</span><span class="p">(</span><span class="mi">5</span><span class="p">),</span> <span class="n">PCA</span><span class="p">(</span><span class="mi">10</span><span class="p">)],</span>
<span class="gp">... </span>                  <span class="n">clf</span><span class="o">=</span><span class="p">[</span><span class="n">SVC</span><span class="p">(),</span> <span class="n">LogisticRegression</span><span class="p">()],</span>
<span class="gp">... </span>                  <span class="n">clf__C</span><span class="o">=</span><span class="p">[</span><span class="mf">0.1</span><span class="p">,</span> <span class="mi">10</span><span class="p">,</span> <span class="mi">100</span><span class="p">])</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">grid_search</span> <span class="o">=</span> <span class="n">GridSearchCV</span><span class="p">(</span><span class="n">pipe</span><span class="p">,</span> <span class="n">param_grid</span><span class="o">=</span><span class="n">param_grid</span><span class="p">)</span>
</pre></div>
</div>
<aside class="topic">
<p class="topic-title">See Also:</p>
<ul class="simple">
<li><p><a class="reference internal" href="grid_search.html#composite-grid-search"><span class="std std-ref">Composite estimators and parameter spaces</span></a></p></li>
</ul>
</aside>
<p></div>
</details></p>
<aside class="topic">
<p class="topic-title">Examples:</p>
<ul class="simple">
<li><p><a class="reference internal" href="../auto_examples/feature_selection/plot_feature_selection_pipeline.html#sphx-glr-auto-examples-feature-selection-plot-feature-selection-pipeline-py"><span class="std std-ref">Pipeline ANOVA SVM</span></a></p></li>
<li><p><a class="reference internal" href="../auto_examples/model_selection/plot_grid_search_text_feature_extraction.html#sphx-glr-auto-examples-model-selection-plot-grid-search-text-feature-extraction-py"><span class="std std-ref">Sample pipeline for text feature extraction and evaluation</span></a></p></li>
<li><p><a class="reference internal" href="../auto_examples/compose/plot_digits_pipe.html#sphx-glr-auto-examples-compose-plot-digits-pipe-py"><span class="std std-ref">Pipelining: chaining a PCA and a logistic regression</span></a></p></li>
<li><p><a class="reference internal" href="../auto_examples/miscellaneous/plot_kernel_approximation.html#sphx-glr-auto-examples-miscellaneous-plot-kernel-approximation-py"><span class="std std-ref">Explicit feature map approximation for RBF kernels</span></a></p></li>
<li><p><a class="reference internal" href="../auto_examples/svm/plot_svm_anova.html#sphx-glr-auto-examples-svm-plot-svm-anova-py"><span class="std std-ref">SVM-Anova: SVM with univariate feature selection</span></a></p></li>
<li><p><a class="reference internal" href="../auto_examples/compose/plot_compare_reduction.html#sphx-glr-auto-examples-compose-plot-compare-reduction-py"><span class="std std-ref">Selecting dimensionality reduction with Pipeline and GridSearchCV</span></a></p></li>
<li><p><a class="reference internal" href="../auto_examples/miscellaneous/plot_pipeline_display.html#sphx-glr-auto-examples-miscellaneous-plot-pipeline-display-py"><span class="std std-ref">Displaying Pipelines</span></a></p></li>
</ul>
</aside>
</section>
</section>
<section id="caching-transformers-avoid-repeated-computation">
<span id="pipeline-cache"></span><h3><span class="section-number">6.1.1.2. </span>Caching transformers: avoid repeated computation<a class="headerlink" href="#caching-transformers-avoid-repeated-computation" title="Link to this heading">¶</a></h3>
<p>Fitting transformers may be computationally expensive. With its
<code class="docutils literal notranslate"><span class="pre">memory</span></code> parameter set, <a class="reference internal" href="generated/sklearn.pipeline.Pipeline.html#sklearn.pipeline.Pipeline" title="sklearn.pipeline.Pipeline"><code class="xref py py-class docutils literal notranslate"><span class="pre">Pipeline</span></code></a> will cache each transformer
after calling <code class="docutils literal notranslate"><span class="pre">fit</span></code>.
This feature is used to avoid computing the fit transformers within a pipeline
if the parameters and input data are identical. A typical example is the case of
a grid search in which the transformers can be fitted only once and reused for
each configuration. The last step will never be cached, even if it is a transformer.</p>
<p>The parameter <code class="docutils literal notranslate"><span class="pre">memory</span></code> is needed in order to cache the transformers.
<code class="docutils literal notranslate"><span class="pre">memory</span></code> can be either a string containing the directory where to cache the
transformers or a <a class="reference external" href="https://joblib.readthedocs.io/en/latest/memory.html">joblib.Memory</a>
object:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">tempfile</span> <span class="kn">import</span> <span class="n">mkdtemp</span>
<span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">shutil</span> <span class="kn">import</span> <span class="n">rmtree</span>
<span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.decomposition</span> <span class="kn">import</span> <span class="n">PCA</span>
<span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.svm</span> <span class="kn">import</span> <span class="n">SVC</span>
<span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.pipeline</span> <span class="kn">import</span> <span class="n">Pipeline</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">estimators</span> <span class="o">=</span> <span class="p">[(</span><span class="s1">&#39;reduce_dim&#39;</span><span class="p">,</span> <span class="n">PCA</span><span class="p">()),</span> <span class="p">(</span><span class="s1">&#39;clf&#39;</span><span class="p">,</span> <span class="n">SVC</span><span class="p">())]</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">cachedir</span> <span class="o">=</span> <span class="n">mkdtemp</span><span class="p">()</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">pipe</span> <span class="o">=</span> <span class="n">Pipeline</span><span class="p">(</span><span class="n">estimators</span><span class="p">,</span> <span class="n">memory</span><span class="o">=</span><span class="n">cachedir</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">pipe</span>
<span class="go">Pipeline(memory=...,</span>
<span class="go">         steps=[(&#39;reduce_dim&#39;, PCA()), (&#39;clf&#39;, SVC())])</span>
<span class="gp">&gt;&gt;&gt; </span><span class="c1"># Clear the cache directory when you don&#39;t need it anymore</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">rmtree</span><span class="p">(</span><span class="n">cachedir</span><span class="p">)</span>
</pre></div>
</div>
<p><details id="summary-anchor">
<summary class="btn btn-light">
<strong>Warning: Side effect of caching transformers</strong>
<span class="tooltiptext">Click for more details</span>
<a class="headerlink" href="#summary-anchor" title="Permalink to this heading">¶</a>
</summary>
<div class="card"></p>
<p>Using a <a class="reference internal" href="generated/sklearn.pipeline.Pipeline.html#sklearn.pipeline.Pipeline" title="sklearn.pipeline.Pipeline"><code class="xref py py-class docutils literal notranslate"><span class="pre">Pipeline</span></code></a> without cache enabled, it is possible to
inspect the original instance such as:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.datasets</span> <span class="kn">import</span> <span class="n">load_digits</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">X_digits</span><span class="p">,</span> <span class="n">y_digits</span> <span class="o">=</span> <span class="n">load_digits</span><span class="p">(</span><span class="n">return_X_y</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">pca1</span> <span class="o">=</span> <span class="n">PCA</span><span class="p">(</span><span class="n">n_components</span><span class="o">=</span><span class="mi">10</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">svm1</span> <span class="o">=</span> <span class="n">SVC</span><span class="p">()</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">pipe</span> <span class="o">=</span> <span class="n">Pipeline</span><span class="p">([(</span><span class="s1">&#39;reduce_dim&#39;</span><span class="p">,</span> <span class="n">pca1</span><span class="p">),</span> <span class="p">(</span><span class="s1">&#39;clf&#39;</span><span class="p">,</span> <span class="n">svm1</span><span class="p">)])</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">pipe</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X_digits</span><span class="p">,</span> <span class="n">y_digits</span><span class="p">)</span>
<span class="go">Pipeline(steps=[(&#39;reduce_dim&#39;, PCA(n_components=10)), (&#39;clf&#39;, SVC())])</span>
<span class="gp">&gt;&gt;&gt; </span><span class="c1"># The pca instance can be inspected directly</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">pca1</span><span class="o">.</span><span class="n">components_</span><span class="o">.</span><span class="n">shape</span>
<span class="go">(10, 64)</span>
</pre></div>
</div>
<p>Enabling caching triggers a clone of the transformers before fitting.
Therefore, the transformer instance given to the pipeline cannot be
inspected directly.
In following example, accessing the <a class="reference internal" href="generated/sklearn.decomposition.PCA.html#sklearn.decomposition.PCA" title="sklearn.decomposition.PCA"><code class="xref py py-class docutils literal notranslate"><span class="pre">PCA</span></code></a>
instance <code class="docutils literal notranslate"><span class="pre">pca2</span></code> will raise an <code class="docutils literal notranslate"><span class="pre">AttributeError</span></code> since <code class="docutils literal notranslate"><span class="pre">pca2</span></code> will be an
unfitted transformer.
Instead, use the attribute <code class="docutils literal notranslate"><span class="pre">named_steps</span></code> to inspect estimators within
the pipeline:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">cachedir</span> <span class="o">=</span> <span class="n">mkdtemp</span><span class="p">()</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">pca2</span> <span class="o">=</span> <span class="n">PCA</span><span class="p">(</span><span class="n">n_components</span><span class="o">=</span><span class="mi">10</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">svm2</span> <span class="o">=</span> <span class="n">SVC</span><span class="p">()</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">cached_pipe</span> <span class="o">=</span> <span class="n">Pipeline</span><span class="p">([(</span><span class="s1">&#39;reduce_dim&#39;</span><span class="p">,</span> <span class="n">pca2</span><span class="p">),</span> <span class="p">(</span><span class="s1">&#39;clf&#39;</span><span class="p">,</span> <span class="n">svm2</span><span class="p">)],</span>
<span class="gp">... </span>                       <span class="n">memory</span><span class="o">=</span><span class="n">cachedir</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">cached_pipe</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X_digits</span><span class="p">,</span> <span class="n">y_digits</span><span class="p">)</span>
<span class="go">Pipeline(memory=...,</span>
<span class="go">         steps=[(&#39;reduce_dim&#39;, PCA(n_components=10)), (&#39;clf&#39;, SVC())])</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">cached_pipe</span><span class="o">.</span><span class="n">named_steps</span><span class="p">[</span><span class="s1">&#39;reduce_dim&#39;</span><span class="p">]</span><span class="o">.</span><span class="n">components_</span><span class="o">.</span><span class="n">shape</span>
<span class="go">(10, 64)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="c1"># Remove the cache directory</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">rmtree</span><span class="p">(</span><span class="n">cachedir</span><span class="p">)</span>
</pre></div>
</div>
<p></div>
</details></p>
<aside class="topic">
<p class="topic-title">Examples:</p>
<ul class="simple">
<li><p><a class="reference internal" href="../auto_examples/compose/plot_compare_reduction.html#sphx-glr-auto-examples-compose-plot-compare-reduction-py"><span class="std std-ref">Selecting dimensionality reduction with Pipeline and GridSearchCV</span></a></p></li>
</ul>
</aside>
</section>
</section>
<section id="transforming-target-in-regression">
<span id="transformed-target-regressor"></span><h2><span class="section-number">6.1.2. </span>Transforming target in regression<a class="headerlink" href="#transforming-target-in-regression" title="Link to this heading">¶</a></h2>
<p><a class="reference internal" href="generated/sklearn.compose.TransformedTargetRegressor.html#sklearn.compose.TransformedTargetRegressor" title="sklearn.compose.TransformedTargetRegressor"><code class="xref py py-class docutils literal notranslate"><span class="pre">TransformedTargetRegressor</span></code></a> transforms the
targets <code class="docutils literal notranslate"><span class="pre">y</span></code> before fitting a regression model. The predictions are mapped
back to the original space via an inverse transform. It takes as an argument
the regressor that will be used for prediction, and the transformer that will
be applied to the target variable:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="kn">import</span> <span class="nn">numpy</span> <span class="k">as</span> <span class="nn">np</span>
<span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.datasets</span> <span class="kn">import</span> <span class="n">fetch_california_housing</span>
<span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.compose</span> <span class="kn">import</span> <span class="n">TransformedTargetRegressor</span>
<span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.preprocessing</span> <span class="kn">import</span> <span class="n">QuantileTransformer</span>
<span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.linear_model</span> <span class="kn">import</span> <span class="n">LinearRegression</span>
<span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.model_selection</span> <span class="kn">import</span> <span class="n">train_test_split</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">X</span><span class="p">,</span> <span class="n">y</span> <span class="o">=</span> <span class="n">fetch_california_housing</span><span class="p">(</span><span class="n">return_X_y</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">X</span><span class="p">,</span> <span class="n">y</span> <span class="o">=</span> <span class="n">X</span><span class="p">[:</span><span class="mi">2000</span><span class="p">,</span> <span class="p">:],</span> <span class="n">y</span><span class="p">[:</span><span class="mi">2000</span><span class="p">]</span>  <span class="c1"># select a subset of data</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">transformer</span> <span class="o">=</span> <span class="n">QuantileTransformer</span><span class="p">(</span><span class="n">output_distribution</span><span class="o">=</span><span class="s1">&#39;normal&#39;</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">regressor</span> <span class="o">=</span> <span class="n">LinearRegression</span><span class="p">()</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">regr</span> <span class="o">=</span> <span class="n">TransformedTargetRegressor</span><span class="p">(</span><span class="n">regressor</span><span class="o">=</span><span class="n">regressor</span><span class="p">,</span>
<span class="gp">... </span>                                  <span class="n">transformer</span><span class="o">=</span><span class="n">transformer</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">X_train</span><span class="p">,</span> <span class="n">X_test</span><span class="p">,</span> <span class="n">y_train</span><span class="p">,</span> <span class="n">y_test</span> <span class="o">=</span> <span class="n">train_test_split</span><span class="p">(</span><span class="n">X</span><span class="p">,</span> <span class="n">y</span><span class="p">,</span> <span class="n">random_state</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">regr</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X_train</span><span class="p">,</span> <span class="n">y_train</span><span class="p">)</span>
<span class="go">TransformedTargetRegressor(...)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="nb">print</span><span class="p">(</span><span class="s1">&#39;R2 score: </span><span class="si">{0:.2f}</span><span class="s1">&#39;</span><span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="n">regr</span><span class="o">.</span><span class="n">score</span><span class="p">(</span><span class="n">X_test</span><span class="p">,</span> <span class="n">y_test</span><span class="p">)))</span>
<span class="go">R2 score: 0.61</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">raw_target_regr</span> <span class="o">=</span> <span class="n">LinearRegression</span><span class="p">()</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X_train</span><span class="p">,</span> <span class="n">y_train</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="nb">print</span><span class="p">(</span><span class="s1">&#39;R2 score: </span><span class="si">{0:.2f}</span><span class="s1">&#39;</span><span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="n">raw_target_regr</span><span class="o">.</span><span class="n">score</span><span class="p">(</span><span class="n">X_test</span><span class="p">,</span> <span class="n">y_test</span><span class="p">)))</span>
<span class="go">R2 score: 0.59</span>
</pre></div>
</div>
<p>For simple transformations, instead of a Transformer object, a pair of
functions can be passed, defining the transformation and its inverse mapping:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="k">def</span> <span class="nf">func</span><span class="p">(</span><span class="n">x</span><span class="p">):</span>
<span class="gp">... </span>    <span class="k">return</span> <span class="n">np</span><span class="o">.</span><span class="n">log</span><span class="p">(</span><span class="n">x</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="k">def</span> <span class="nf">inverse_func</span><span class="p">(</span><span class="n">x</span><span class="p">):</span>
<span class="gp">... </span>    <span class="k">return</span> <span class="n">np</span><span class="o">.</span><span class="n">exp</span><span class="p">(</span><span class="n">x</span><span class="p">)</span>
</pre></div>
</div>
<p>Subsequently, the object is created as:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">regr</span> <span class="o">=</span> <span class="n">TransformedTargetRegressor</span><span class="p">(</span><span class="n">regressor</span><span class="o">=</span><span class="n">regressor</span><span class="p">,</span>
<span class="gp">... </span>                                  <span class="n">func</span><span class="o">=</span><span class="n">func</span><span class="p">,</span>
<span class="gp">... </span>                                  <span class="n">inverse_func</span><span class="o">=</span><span class="n">inverse_func</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">regr</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X_train</span><span class="p">,</span> <span class="n">y_train</span><span class="p">)</span>
<span class="go">TransformedTargetRegressor(...)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="nb">print</span><span class="p">(</span><span class="s1">&#39;R2 score: </span><span class="si">{0:.2f}</span><span class="s1">&#39;</span><span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="n">regr</span><span class="o">.</span><span class="n">score</span><span class="p">(</span><span class="n">X_test</span><span class="p">,</span> <span class="n">y_test</span><span class="p">)))</span>
<span class="go">R2 score: 0.51</span>
</pre></div>
</div>
<p>By default, the provided functions are checked at each fit to be the inverse of
each other. However, it is possible to bypass this checking by setting
<code class="docutils literal notranslate"><span class="pre">check_inverse</span></code> to <code class="docutils literal notranslate"><span class="pre">False</span></code>:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="k">def</span> <span class="nf">inverse_func</span><span class="p">(</span><span class="n">x</span><span class="p">):</span>
<span class="gp">... </span>    <span class="k">return</span> <span class="n">x</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">regr</span> <span class="o">=</span> <span class="n">TransformedTargetRegressor</span><span class="p">(</span><span class="n">regressor</span><span class="o">=</span><span class="n">regressor</span><span class="p">,</span>
<span class="gp">... </span>                                  <span class="n">func</span><span class="o">=</span><span class="n">func</span><span class="p">,</span>
<span class="gp">... </span>                                  <span class="n">inverse_func</span><span class="o">=</span><span class="n">inverse_func</span><span class="p">,</span>
<span class="gp">... </span>                                  <span class="n">check_inverse</span><span class="o">=</span><span class="kc">False</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">regr</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X_train</span><span class="p">,</span> <span class="n">y_train</span><span class="p">)</span>
<span class="go">TransformedTargetRegressor(...)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="nb">print</span><span class="p">(</span><span class="s1">&#39;R2 score: </span><span class="si">{0:.2f}</span><span class="s1">&#39;</span><span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="n">regr</span><span class="o">.</span><span class="n">score</span><span class="p">(</span><span class="n">X_test</span><span class="p">,</span> <span class="n">y_test</span><span class="p">)))</span>
<span class="go">R2 score: -1.57</span>
</pre></div>
</div>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>The transformation can be triggered by setting either <code class="docutils literal notranslate"><span class="pre">transformer</span></code> or the
pair of functions <code class="docutils literal notranslate"><span class="pre">func</span></code> and <code class="docutils literal notranslate"><span class="pre">inverse_func</span></code>. However, setting both
options will raise an error.</p>
</div>
<aside class="topic">
<p class="topic-title">Examples:</p>
<ul class="simple">
<li><p><a class="reference internal" href="../auto_examples/compose/plot_transformed_target.html#sphx-glr-auto-examples-compose-plot-transformed-target-py"><span class="std std-ref">Effect of transforming the targets in regression model</span></a></p></li>
</ul>
</aside>
</section>
<section id="featureunion-composite-feature-spaces">
<span id="feature-union"></span><h2><span class="section-number">6.1.3. </span>FeatureUnion: composite feature spaces<a class="headerlink" href="#featureunion-composite-feature-spaces" title="Link to this heading">¶</a></h2>
<p><a class="reference internal" href="generated/sklearn.pipeline.FeatureUnion.html#sklearn.pipeline.FeatureUnion" title="sklearn.pipeline.FeatureUnion"><code class="xref py py-class docutils literal notranslate"><span class="pre">FeatureUnion</span></code></a> combines several transformer objects into a new
transformer that combines their output. A <a class="reference internal" href="generated/sklearn.pipeline.FeatureUnion.html#sklearn.pipeline.FeatureUnion" title="sklearn.pipeline.FeatureUnion"><code class="xref py py-class docutils literal notranslate"><span class="pre">FeatureUnion</span></code></a> takes
a list of transformer objects. During fitting, each of these
is fit to the data independently. The transformers are applied in parallel,
and the feature matrices they output are concatenated side-by-side into a
larger matrix.</p>
<p>When you want to apply different transformations to each field of the data,
see the related class <a class="reference internal" href="generated/sklearn.compose.ColumnTransformer.html#sklearn.compose.ColumnTransformer" title="sklearn.compose.ColumnTransformer"><code class="xref py py-class docutils literal notranslate"><span class="pre">ColumnTransformer</span></code></a>
(see <a class="reference internal" href="#column-transformer"><span class="std std-ref">user guide</span></a>).</p>
<p><a class="reference internal" href="generated/sklearn.pipeline.FeatureUnion.html#sklearn.pipeline.FeatureUnion" title="sklearn.pipeline.FeatureUnion"><code class="xref py py-class docutils literal notranslate"><span class="pre">FeatureUnion</span></code></a> serves the same purposes as <a class="reference internal" href="generated/sklearn.pipeline.Pipeline.html#sklearn.pipeline.Pipeline" title="sklearn.pipeline.Pipeline"><code class="xref py py-class docutils literal notranslate"><span class="pre">Pipeline</span></code></a> -
convenience and joint parameter estimation and validation.</p>
<p><a class="reference internal" href="generated/sklearn.pipeline.FeatureUnion.html#sklearn.pipeline.FeatureUnion" title="sklearn.pipeline.FeatureUnion"><code class="xref py py-class docutils literal notranslate"><span class="pre">FeatureUnion</span></code></a> and <a class="reference internal" href="generated/sklearn.pipeline.Pipeline.html#sklearn.pipeline.Pipeline" title="sklearn.pipeline.Pipeline"><code class="xref py py-class docutils literal notranslate"><span class="pre">Pipeline</span></code></a> can be combined to
create complex models.</p>
<p>(A <a class="reference internal" href="generated/sklearn.pipeline.FeatureUnion.html#sklearn.pipeline.FeatureUnion" title="sklearn.pipeline.FeatureUnion"><code class="xref py py-class docutils literal notranslate"><span class="pre">FeatureUnion</span></code></a> has no way of checking whether two transformers
might produce identical features. It only produces a union when the
feature sets are disjoint, and making sure they are is the caller’s
responsibility.)</p>
<section id="id1">
<h3><span class="section-number">6.1.3.1. </span>Usage<a class="headerlink" href="#id1" title="Link to this heading">¶</a></h3>
<p>A <a class="reference internal" href="generated/sklearn.pipeline.FeatureUnion.html#sklearn.pipeline.FeatureUnion" title="sklearn.pipeline.FeatureUnion"><code class="xref py py-class docutils literal notranslate"><span class="pre">FeatureUnion</span></code></a> is built using a list of <code class="docutils literal notranslate"><span class="pre">(key,</span> <span class="pre">value)</span></code> pairs,
where the <code class="docutils literal notranslate"><span class="pre">key</span></code> is the name you want to give to a given transformation
(an arbitrary string; it only serves as an identifier)
and <code class="docutils literal notranslate"><span class="pre">value</span></code> is an estimator object:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.pipeline</span> <span class="kn">import</span> <span class="n">FeatureUnion</span>
<span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.decomposition</span> <span class="kn">import</span> <span class="n">PCA</span>
<span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.decomposition</span> <span class="kn">import</span> <span class="n">KernelPCA</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">estimators</span> <span class="o">=</span> <span class="p">[(</span><span class="s1">&#39;linear_pca&#39;</span><span class="p">,</span> <span class="n">PCA</span><span class="p">()),</span> <span class="p">(</span><span class="s1">&#39;kernel_pca&#39;</span><span class="p">,</span> <span class="n">KernelPCA</span><span class="p">())]</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">combined</span> <span class="o">=</span> <span class="n">FeatureUnion</span><span class="p">(</span><span class="n">estimators</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">combined</span>
<span class="go">FeatureUnion(transformer_list=[(&#39;linear_pca&#39;, PCA()),</span>
<span class="go">                               (&#39;kernel_pca&#39;, KernelPCA())])</span>
</pre></div>
</div>
<p>Like pipelines, feature unions have a shorthand constructor called
<a class="reference internal" href="generated/sklearn.pipeline.make_union.html#sklearn.pipeline.make_union" title="sklearn.pipeline.make_union"><code class="xref py py-func docutils literal notranslate"><span class="pre">make_union</span></code></a> that does not require explicit naming of the components.</p>
<p>Like <code class="docutils literal notranslate"><span class="pre">Pipeline</span></code>, individual steps may be replaced using <code class="docutils literal notranslate"><span class="pre">set_params</span></code>,
and ignored by setting to <code class="docutils literal notranslate"><span class="pre">'drop'</span></code>:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">combined</span><span class="o">.</span><span class="n">set_params</span><span class="p">(</span><span class="n">kernel_pca</span><span class="o">=</span><span class="s1">&#39;drop&#39;</span><span class="p">)</span>
<span class="go">FeatureUnion(transformer_list=[(&#39;linear_pca&#39;, PCA()),</span>
<span class="go">                               (&#39;kernel_pca&#39;, &#39;drop&#39;)])</span>
</pre></div>
</div>
<aside class="topic">
<p class="topic-title">Examples:</p>
<ul class="simple">
<li><p><a class="reference internal" href="../auto_examples/compose/plot_feature_union.html#sphx-glr-auto-examples-compose-plot-feature-union-py"><span class="std std-ref">Concatenating multiple feature extraction methods</span></a></p></li>
</ul>
</aside>
</section>
</section>
<section id="columntransformer-for-heterogeneous-data">
<span id="column-transformer"></span><h2><span class="section-number">6.1.4. </span>ColumnTransformer for heterogeneous data<a class="headerlink" href="#columntransformer-for-heterogeneous-data" title="Link to this heading">¶</a></h2>
<p>Many datasets contain features of different types, say text, floats, and dates,
where each type of feature requires separate preprocessing or feature
extraction steps.  Often it is easiest to preprocess data before applying
scikit-learn methods, for example using <a class="reference external" href="https://pandas.pydata.org/">pandas</a>.
Processing your data before passing it to scikit-learn might be problematic for
one of the following reasons:</p>
<ol class="arabic simple">
<li><p>Incorporating statistics from test data into the preprocessors makes
cross-validation scores unreliable (known as <em>data leakage</em>),
for example in the case of scalers or imputing missing values.</p></li>
<li><p>You may want to include the parameters of the preprocessors in a
<a class="reference internal" href="grid_search.html#grid-search"><span class="std std-ref">parameter search</span></a>.</p></li>
</ol>
<p>The <a class="reference internal" href="generated/sklearn.compose.ColumnTransformer.html#sklearn.compose.ColumnTransformer" title="sklearn.compose.ColumnTransformer"><code class="xref py py-class docutils literal notranslate"><span class="pre">ColumnTransformer</span></code></a> helps performing different
transformations for different columns of the data, within a
<a class="reference internal" href="generated/sklearn.pipeline.Pipeline.html#sklearn.pipeline.Pipeline" title="sklearn.pipeline.Pipeline"><code class="xref py py-class docutils literal notranslate"><span class="pre">Pipeline</span></code></a> that is safe from data leakage and that can
be parametrized. <a class="reference internal" href="generated/sklearn.compose.ColumnTransformer.html#sklearn.compose.ColumnTransformer" title="sklearn.compose.ColumnTransformer"><code class="xref py py-class docutils literal notranslate"><span class="pre">ColumnTransformer</span></code></a> works on
arrays, sparse matrices, and
<a class="reference external" href="https://pandas.pydata.org/pandas-docs/stable/">pandas DataFrames</a>.</p>
<p>To each column, a different transformation can be applied, such as
preprocessing or a specific feature extraction method:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="kn">import</span> <span class="nn">pandas</span> <span class="k">as</span> <span class="nn">pd</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">X</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">DataFrame</span><span class="p">(</span>
<span class="gp">... </span>    <span class="p">{</span><span class="s1">&#39;city&#39;</span><span class="p">:</span> <span class="p">[</span><span class="s1">&#39;London&#39;</span><span class="p">,</span> <span class="s1">&#39;London&#39;</span><span class="p">,</span> <span class="s1">&#39;Paris&#39;</span><span class="p">,</span> <span class="s1">&#39;Sallisaw&#39;</span><span class="p">],</span>
<span class="gp">... </span>     <span class="s1">&#39;title&#39;</span><span class="p">:</span> <span class="p">[</span><span class="s2">&quot;His Last Bow&quot;</span><span class="p">,</span> <span class="s2">&quot;How Watson Learned the Trick&quot;</span><span class="p">,</span>
<span class="gp">... </span>               <span class="s2">&quot;A Moveable Feast&quot;</span><span class="p">,</span> <span class="s2">&quot;The Grapes of Wrath&quot;</span><span class="p">],</span>
<span class="gp">... </span>     <span class="s1">&#39;expert_rating&#39;</span><span class="p">:</span> <span class="p">[</span><span class="mi">5</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">,</span> <span class="mi">5</span><span class="p">],</span>
<span class="gp">... </span>     <span class="s1">&#39;user_rating&#39;</span><span class="p">:</span> <span class="p">[</span><span class="mi">4</span><span class="p">,</span> <span class="mi">5</span><span class="p">,</span> <span class="mi">4</span><span class="p">,</span> <span class="mi">3</span><span class="p">]})</span>
</pre></div>
</div>
<p>For this data, we might want to encode the <code class="docutils literal notranslate"><span class="pre">'city'</span></code> column as a categorical
variable using <a class="reference internal" href="generated/sklearn.preprocessing.OneHotEncoder.html#sklearn.preprocessing.OneHotEncoder" title="sklearn.preprocessing.OneHotEncoder"><code class="xref py py-class docutils literal notranslate"><span class="pre">OneHotEncoder</span></code></a> but apply a
<a class="reference internal" href="generated/sklearn.feature_extraction.text.CountVectorizer.html#sklearn.feature_extraction.text.CountVectorizer" title="sklearn.feature_extraction.text.CountVectorizer"><code class="xref py py-class docutils literal notranslate"><span class="pre">CountVectorizer</span></code></a> to the <code class="docutils literal notranslate"><span class="pre">'title'</span></code> column.
As we might use multiple feature extraction methods on the same column, we give
each transformer a unique name, say <code class="docutils literal notranslate"><span class="pre">'city_category'</span></code> and <code class="docutils literal notranslate"><span class="pre">'title_bow'</span></code>.
By default, the remaining rating columns are ignored (<code class="docutils literal notranslate"><span class="pre">remainder='drop'</span></code>):</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.compose</span> <span class="kn">import</span> <span class="n">ColumnTransformer</span>
<span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.feature_extraction.text</span> <span class="kn">import</span> <span class="n">CountVectorizer</span>
<span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.preprocessing</span> <span class="kn">import</span> <span class="n">OneHotEncoder</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">column_trans</span> <span class="o">=</span> <span class="n">ColumnTransformer</span><span class="p">(</span>
<span class="gp">... </span>    <span class="p">[(</span><span class="s1">&#39;categories&#39;</span><span class="p">,</span> <span class="n">OneHotEncoder</span><span class="p">(</span><span class="n">dtype</span><span class="o">=</span><span class="s1">&#39;int&#39;</span><span class="p">),</span> <span class="p">[</span><span class="s1">&#39;city&#39;</span><span class="p">]),</span>
<span class="gp">... </span>     <span class="p">(</span><span class="s1">&#39;title_bow&#39;</span><span class="p">,</span> <span class="n">CountVectorizer</span><span class="p">(),</span> <span class="s1">&#39;title&#39;</span><span class="p">)],</span>
<span class="gp">... </span>    <span class="n">remainder</span><span class="o">=</span><span class="s1">&#39;drop&#39;</span><span class="p">,</span> <span class="n">verbose_feature_names_out</span><span class="o">=</span><span class="kc">False</span><span class="p">)</span>

<span class="gp">&gt;&gt;&gt; </span><span class="n">column_trans</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X</span><span class="p">)</span>
<span class="go">ColumnTransformer(transformers=[(&#39;categories&#39;, OneHotEncoder(dtype=&#39;int&#39;),</span>
<span class="go">                                 [&#39;city&#39;]),</span>
<span class="go">                                (&#39;title_bow&#39;, CountVectorizer(), &#39;title&#39;)],</span>
<span class="go">                  verbose_feature_names_out=False)</span>

<span class="gp">&gt;&gt;&gt; </span><span class="n">column_trans</span><span class="o">.</span><span class="n">get_feature_names_out</span><span class="p">()</span>
<span class="go">array([&#39;city_London&#39;, &#39;city_Paris&#39;, &#39;city_Sallisaw&#39;, &#39;bow&#39;, &#39;feast&#39;,</span>
<span class="go">&#39;grapes&#39;, &#39;his&#39;, &#39;how&#39;, &#39;last&#39;, &#39;learned&#39;, &#39;moveable&#39;, &#39;of&#39;, &#39;the&#39;,</span>
<span class="go"> &#39;trick&#39;, &#39;watson&#39;, &#39;wrath&#39;], ...)</span>

<span class="gp">&gt;&gt;&gt; </span><span class="n">column_trans</span><span class="o">.</span><span class="n">transform</span><span class="p">(</span><span class="n">X</span><span class="p">)</span><span class="o">.</span><span class="n">toarray</span><span class="p">()</span>
<span class="go">array([[1, 0, 0, 1, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0],</span>
<span class="go">       [1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 1, 1, 1, 0],</span>
<span class="go">       [0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0],</span>
<span class="go">       [0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1]]...)</span>
</pre></div>
</div>
<p>In the above example, the
<a class="reference internal" href="generated/sklearn.feature_extraction.text.CountVectorizer.html#sklearn.feature_extraction.text.CountVectorizer" title="sklearn.feature_extraction.text.CountVectorizer"><code class="xref py py-class docutils literal notranslate"><span class="pre">CountVectorizer</span></code></a> expects a 1D array as
input and therefore the columns were specified as a string (<code class="docutils literal notranslate"><span class="pre">'title'</span></code>).
However, <a class="reference internal" href="generated/sklearn.preprocessing.OneHotEncoder.html#sklearn.preprocessing.OneHotEncoder" title="sklearn.preprocessing.OneHotEncoder"><code class="xref py py-class docutils literal notranslate"><span class="pre">OneHotEncoder</span></code></a>
as most of other transformers expects 2D data, therefore in that case you need
to specify the column as a list of strings (<code class="docutils literal notranslate"><span class="pre">['city']</span></code>).</p>
<p>Apart from a scalar or a single item list, the column selection can be specified
as a list of multiple items, an integer array, a slice, a boolean mask, or
with a <a class="reference internal" href="generated/sklearn.compose.make_column_selector.html#sklearn.compose.make_column_selector" title="sklearn.compose.make_column_selector"><code class="xref py py-func docutils literal notranslate"><span class="pre">make_column_selector</span></code></a>. The
<a class="reference internal" href="generated/sklearn.compose.make_column_selector.html#sklearn.compose.make_column_selector" title="sklearn.compose.make_column_selector"><code class="xref py py-func docutils literal notranslate"><span class="pre">make_column_selector</span></code></a> is used to select columns based
on data type or column name:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.preprocessing</span> <span class="kn">import</span> <span class="n">StandardScaler</span>
<span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.compose</span> <span class="kn">import</span> <span class="n">make_column_selector</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">ct</span> <span class="o">=</span> <span class="n">ColumnTransformer</span><span class="p">([</span>
<span class="gp">... </span>      <span class="p">(</span><span class="s1">&#39;scale&#39;</span><span class="p">,</span> <span class="n">StandardScaler</span><span class="p">(),</span>
<span class="gp">... </span>      <span class="n">make_column_selector</span><span class="p">(</span><span class="n">dtype_include</span><span class="o">=</span><span class="n">np</span><span class="o">.</span><span class="n">number</span><span class="p">)),</span>
<span class="gp">... </span>      <span class="p">(</span><span class="s1">&#39;onehot&#39;</span><span class="p">,</span>
<span class="gp">... </span>      <span class="n">OneHotEncoder</span><span class="p">(),</span>
<span class="gp">... </span>      <span class="n">make_column_selector</span><span class="p">(</span><span class="n">pattern</span><span class="o">=</span><span class="s1">&#39;city&#39;</span><span class="p">,</span> <span class="n">dtype_include</span><span class="o">=</span><span class="nb">object</span><span class="p">))])</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">ct</span><span class="o">.</span><span class="n">fit_transform</span><span class="p">(</span><span class="n">X</span><span class="p">)</span>
<span class="go">array([[ 0.904...,  0.      ,  1. ,  0. ,  0. ],</span>
<span class="go">       [-1.507...,  1.414...,  1. ,  0. ,  0. ],</span>
<span class="go">       [-0.301...,  0.      ,  0. ,  1. ,  0. ],</span>
<span class="go">       [ 0.904..., -1.414...,  0. ,  0. ,  1. ]])</span>
</pre></div>
</div>
<p>Strings can reference columns if the input is a DataFrame, integers are always
interpreted as the positional columns.</p>
<p>We can keep the remaining rating columns by setting
<code class="docutils literal notranslate"><span class="pre">remainder='passthrough'</span></code>. The values are appended to the end of the
transformation:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">column_trans</span> <span class="o">=</span> <span class="n">ColumnTransformer</span><span class="p">(</span>
<span class="gp">... </span>    <span class="p">[(</span><span class="s1">&#39;city_category&#39;</span><span class="p">,</span> <span class="n">OneHotEncoder</span><span class="p">(</span><span class="n">dtype</span><span class="o">=</span><span class="s1">&#39;int&#39;</span><span class="p">),[</span><span class="s1">&#39;city&#39;</span><span class="p">]),</span>
<span class="gp">... </span>     <span class="p">(</span><span class="s1">&#39;title_bow&#39;</span><span class="p">,</span> <span class="n">CountVectorizer</span><span class="p">(),</span> <span class="s1">&#39;title&#39;</span><span class="p">)],</span>
<span class="gp">... </span>    <span class="n">remainder</span><span class="o">=</span><span class="s1">&#39;passthrough&#39;</span><span class="p">)</span>

<span class="gp">&gt;&gt;&gt; </span><span class="n">column_trans</span><span class="o">.</span><span class="n">fit_transform</span><span class="p">(</span><span class="n">X</span><span class="p">)</span>
<span class="go">array([[1, 0, 0, 1, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 5, 4],</span>
<span class="go">       [1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 1, 1, 1, 0, 3, 5],</span>
<span class="go">       [0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 4, 4],</span>
<span class="go">       [0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 5, 3]]...)</span>
</pre></div>
</div>
<p>The <code class="docutils literal notranslate"><span class="pre">remainder</span></code> parameter can be set to an estimator to transform the
remaining rating columns. The transformed values are appended to the end of
the transformation:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.preprocessing</span> <span class="kn">import</span> <span class="n">MinMaxScaler</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">column_trans</span> <span class="o">=</span> <span class="n">ColumnTransformer</span><span class="p">(</span>
<span class="gp">... </span>    <span class="p">[(</span><span class="s1">&#39;city_category&#39;</span><span class="p">,</span> <span class="n">OneHotEncoder</span><span class="p">(),</span> <span class="p">[</span><span class="s1">&#39;city&#39;</span><span class="p">]),</span>
<span class="gp">... </span>     <span class="p">(</span><span class="s1">&#39;title_bow&#39;</span><span class="p">,</span> <span class="n">CountVectorizer</span><span class="p">(),</span> <span class="s1">&#39;title&#39;</span><span class="p">)],</span>
<span class="gp">... </span>    <span class="n">remainder</span><span class="o">=</span><span class="n">MinMaxScaler</span><span class="p">())</span>

<span class="gp">&gt;&gt;&gt; </span><span class="n">column_trans</span><span class="o">.</span><span class="n">fit_transform</span><span class="p">(</span><span class="n">X</span><span class="p">)[:,</span> <span class="o">-</span><span class="mi">2</span><span class="p">:]</span>
<span class="go">array([[1. , 0.5],</span>
<span class="go">       [0. , 1. ],</span>
<span class="go">       [0.5, 0.5],</span>
<span class="go">       [1. , 0. ]])</span>
</pre></div>
</div>
<p id="make-column-transformer">The <a class="reference internal" href="generated/sklearn.compose.make_column_transformer.html#sklearn.compose.make_column_transformer" title="sklearn.compose.make_column_transformer"><code class="xref py py-func docutils literal notranslate"><span class="pre">make_column_transformer</span></code></a> function is available
to more easily create a <a class="reference internal" href="generated/sklearn.compose.ColumnTransformer.html#sklearn.compose.ColumnTransformer" title="sklearn.compose.ColumnTransformer"><code class="xref py py-class docutils literal notranslate"><span class="pre">ColumnTransformer</span></code></a> object.
Specifically, the names will be given automatically. The equivalent for the
above example would be:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.compose</span> <span class="kn">import</span> <span class="n">make_column_transformer</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">column_trans</span> <span class="o">=</span> <span class="n">make_column_transformer</span><span class="p">(</span>
<span class="gp">... </span>    <span class="p">(</span><span class="n">OneHotEncoder</span><span class="p">(),</span> <span class="p">[</span><span class="s1">&#39;city&#39;</span><span class="p">]),</span>
<span class="gp">... </span>    <span class="p">(</span><span class="n">CountVectorizer</span><span class="p">(),</span> <span class="s1">&#39;title&#39;</span><span class="p">),</span>
<span class="gp">... </span>    <span class="n">remainder</span><span class="o">=</span><span class="n">MinMaxScaler</span><span class="p">())</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">column_trans</span>
<span class="go">ColumnTransformer(remainder=MinMaxScaler(),</span>
<span class="go">                  transformers=[(&#39;onehotencoder&#39;, OneHotEncoder(), [&#39;city&#39;]),</span>
<span class="go">                                (&#39;countvectorizer&#39;, CountVectorizer(),</span>
<span class="go">                                 &#39;title&#39;)])</span>
</pre></div>
</div>
<p>If <a class="reference internal" href="generated/sklearn.compose.ColumnTransformer.html#sklearn.compose.ColumnTransformer" title="sklearn.compose.ColumnTransformer"><code class="xref py py-class docutils literal notranslate"><span class="pre">ColumnTransformer</span></code></a> is fitted with a dataframe
and the dataframe only has string column names, then transforming a dataframe
will use the column names to select the columns:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">ct</span> <span class="o">=</span> <span class="n">ColumnTransformer</span><span class="p">(</span>
<span class="gp">... </span>         <span class="p">[(</span><span class="s2">&quot;scale&quot;</span><span class="p">,</span> <span class="n">StandardScaler</span><span class="p">(),</span> <span class="p">[</span><span class="s2">&quot;expert_rating&quot;</span><span class="p">])])</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">X_new</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">DataFrame</span><span class="p">({</span><span class="s2">&quot;expert_rating&quot;</span><span class="p">:</span> <span class="p">[</span><span class="mi">5</span><span class="p">,</span> <span class="mi">6</span><span class="p">,</span> <span class="mi">1</span><span class="p">],</span>
<span class="gp">... </span>                      <span class="s2">&quot;ignored_new_col&quot;</span><span class="p">:</span> <span class="p">[</span><span class="mf">1.2</span><span class="p">,</span> <span class="mf">0.3</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.1</span><span class="p">]})</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">ct</span><span class="o">.</span><span class="n">transform</span><span class="p">(</span><span class="n">X_new</span><span class="p">)</span>
<span class="go">array([[ 0.9...],</span>
<span class="go">       [ 2.1...],</span>
<span class="go">       [-3.9...]])</span>
</pre></div>
</div>
</section>
<section id="visualizing-composite-estimators">
<span id="id2"></span><h2><span class="section-number">6.1.5. </span>Visualizing Composite Estimators<a class="headerlink" href="#visualizing-composite-estimators" title="Link to this heading">¶</a></h2>
<p>Estimators are displayed with an HTML representation when shown in a
jupyter notebook. This is useful to diagnose or visualize a Pipeline with
many estimators. This visualization is activated by default:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">column_trans</span>  
</pre></div>
</div>
<p>It can be deactivated by setting the <code class="docutils literal notranslate"><span class="pre">display</span></code> option in <a class="reference internal" href="generated/sklearn.set_config.html#sklearn.set_config" title="sklearn.set_config"><code class="xref py py-func docutils literal notranslate"><span class="pre">set_config</span></code></a>
to ‘text’:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn</span> <span class="kn">import</span> <span class="n">set_config</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">set_config</span><span class="p">(</span><span class="n">display</span><span class="o">=</span><span class="s1">&#39;text&#39;</span><span class="p">)</span>  
<span class="gp">&gt;&gt;&gt; </span><span class="c1"># displays text representation in a jupyter context</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">column_trans</span>  
</pre></div>
</div>
<p>An example of the HTML output can be seen in the
<strong>HTML representation of Pipeline</strong> section of
<a class="reference internal" href="../auto_examples/compose/plot_column_transformer_mixed_types.html#sphx-glr-auto-examples-compose-plot-column-transformer-mixed-types-py"><span class="std std-ref">Column Transformer with Mixed Types</span></a>.
As an alternative, the HTML can be written to a file using
<a class="reference internal" href="generated/sklearn.utils.estimator_html_repr.html#sklearn.utils.estimator_html_repr" title="sklearn.utils.estimator_html_repr"><code class="xref py py-func docutils literal notranslate"><span class="pre">estimator_html_repr</span></code></a>:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="kn">from</span> <span class="nn">sklearn.utils</span> <span class="kn">import</span> <span class="n">estimator_html_repr</span>
<span class="gp">&gt;&gt;&gt; </span><span class="k">with</span> <span class="nb">open</span><span class="p">(</span><span class="s1">&#39;my_estimator.html&#39;</span><span class="p">,</span> <span class="s1">&#39;w&#39;</span><span class="p">)</span> <span class="k">as</span> <span class="n">f</span><span class="p">:</span>  
<span class="gp">... </span>    <span class="n">f</span><span class="o">.</span><span class="n">write</span><span class="p">(</span><span class="n">estimator_html_repr</span><span class="p">(</span><span class="n">clf</span><span class="p">))</span>
</pre></div>
</div>
<aside class="topic">
<p class="topic-title">Examples:</p>
<ul class="simple">
<li><p><a class="reference internal" href="../auto_examples/compose/plot_column_transformer.html#sphx-glr-auto-examples-compose-plot-column-transformer-py"><span class="std std-ref">Column Transformer with Heterogeneous Data Sources</span></a></p></li>
<li><p><a class="reference internal" href="../auto_examples/compose/plot_column_transformer_mixed_types.html#sphx-glr-auto-examples-compose-plot-column-transformer-mixed-types-py"><span class="std std-ref">Column Transformer with Mixed Types</span></a></p></li>
</ul>
</aside>
</section>
</section>


      </div>
    <div class="container">
      <footer class="sk-content-footer">
            &copy; 2007 - 2024, scikit-learn developers (BSD License).
          <a href="../_sources/modules/compose.rst.txt" rel="nofollow">Show this page source</a>
      </footer>
    </div>
  </div>
</div>
<script src="../_static/js/vendor/bootstrap.min.js"></script>

<script>
    window.ga=window.ga||function(){(ga.q=ga.q||[]).push(arguments)};ga.l=+new Date;
    ga('create', 'UA-22606712-2', 'auto');
    ga('set', 'anonymizeIp', true);
    ga('send', 'pageview');
</script>
<script async src='https://www.google-analytics.com/analytics.js'></script>


<script defer data-domain="scikit-learn.org" src="https://views.scientific-python.org/js/script.js">
</script>


<script src="../_static/clipboard.min.js"></script>
<script src="../_static/copybutton.js"></script>

<script>
$(document).ready(function() {
    /* Add a [>>>] button on the top-right corner of code samples to hide
     * the >>> and ... prompts and the output and thus make the code
     * copyable. */
    var div = $('.highlight-python .highlight,' +
                '.highlight-python3 .highlight,' +
                '.highlight-pycon .highlight,' +
		'.highlight-default .highlight')
    var pre = div.find('pre');

    // get the styles from the current theme
    pre.parent().parent().css('position', 'relative');

    // create and add the button to all the code blocks that contain >>>
    div.each(function(index) {
        var jthis = $(this);
        // tracebacks (.gt) contain bare text elements that need to be
        // wrapped in a span to work with .nextUntil() (see later)
        jthis.find('pre:has(.gt)').contents().filter(function() {
            return ((this.nodeType == 3) && (this.data.trim().length > 0));
        }).wrap('<span>');
    });

	/*** Add permalink buttons next to glossary terms ***/
	$('dl.glossary > dt[id]').append(function() {
		return ('<a class="headerlink" href="#' +
			    this.getAttribute('id') +
			    '" title="Permalink to this term">¶</a>');
	});
});

</script>
    
<script id="MathJax-script" async src="https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-chtml.js"></script>
    
    <script src="https://scikit-learn.org/versionwarning.js"></script>
</body>
</html>