dev/model_persistence.html



<!DOCTYPE html>
<!--[if IE 8]><html class="no-js lt-ie9" lang="en" > <![endif]-->
<!--[if gt IE 8]><!--> <html class="no-js" lang="en" > <!--<![endif]-->
<head>
  <meta charset="utf-8">
  <meta name="generator" content="Docutils 0.19: https://docutils.sourceforge.io/" />
<meta property="og:title" content="9. Model persistence" />
<meta property="og:type" content="website" />
<meta property="og:url" content="https://scikit-learn/stable/model_persistence.html" />
<meta property="og:site_name" content="scikit-learn" />
<meta property="og:description" content="After training a scikit-learn model, it is desirable to have a way to persist the model for future use without having to retrain. The following sections give you some hints on how to persist a scik..." />
<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="After training a scikit-learn model, it is desirable to have a way to persist the model for future use without having to retrain. The following sections give you some hints on how to persist a scik..." />

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

  
  <title>9. Model persistence &mdash; scikit-learn 1.3.dev0 documentation</title>
  
  <link rel="canonical" href="http://scikit-learn.org/stable/model_persistence.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/plot_directive.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> 
</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="modules/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/v1.3.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/v1.3.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="sk-sidebar-toc-logo">
        </div>
        <div class="btn-group w-100 mb-2" role="group" aria-label="rellinks">
            <a href="computing/parallelism.html" role="button" class="btn sk-btn-rellink py-1" sk-rellink-tooltip="8.3. Parallelism, resource management, and configuration">Prev</a><a href="user_guide.html" role="button" class="btn sk-btn-rellink py-1" sk-rellink-tooltip="User Guide">Up</a>
            <a href="common_pitfalls.html" role="button" class="btn sk-btn-rellink py-1" sk-rellink-tooltip="10. Common pitfalls and recommended practices">Next</a>
        </div>
        <div class="alert alert-danger p-1 mb-2" role="alert">
          <p class="text-center mb-0">
          <strong>scikit-learn 1.3.dev0</strong><br/>
          <a href="http://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 href="user_guide.html" class="sk-toc-active">User Guide</a>
              </li>
              <ul>
              
                <li>
                  <a href="supervised_learning.html" class="">1. Supervised learning</a>
                  
                </li>
              
                <li>
                  <a href="unsupervised_learning.html" class="">2. Unsupervised learning</a>
                  
                </li>
              
                <li>
                  <a href="model_selection.html" class="">3. Model selection and evaluation</a>
                  
                </li>
              
                <li>
                  <a href="inspection.html" class="">4. Inspection</a>
                  
                </li>
              
                <li>
                  <a href="visualizations.html" class="">5. Visualizations</a>
                  
                </li>
              
                <li>
                  <a href="data_transforms.html" class="">6. Dataset transformations</a>
                  
                </li>
              
                <li>
                  <a href="datasets.html" class="">7. Dataset loading utilities</a>
                  
                </li>
              
                <li>
                  <a href="computing.html" class="">8. Computing with scikit-learn</a>
                  
                </li>
              
                <li>
                  <a href="" class="sk-toc-active">9. Model persistence</a>
                  
                </li>
              
                <li>
                  <a href="common_pitfalls.html" class="">10. Common pitfalls and recommended practices</a>
                  
                </li>
              
                <li>
                  <a href="dispatching.html" class="">11. Dispatching</a>
                  
                </li>
              
              </ul>
              
              
              </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="model-persistence">
<span id="id1"></span><h1><span class="section-number">9. </span>Model persistence<a class="headerlink" href="#model-persistence" title="Permalink to this heading">¶</a></h1>
<p>After training a scikit-learn model, it is desirable to have a way to persist
the model for future use without having to retrain. The following sections give
you some hints on how to persist a scikit-learn model.</p>
<section id="python-specific-serialization">
<h2><span class="section-number">9.1. </span>Python specific serialization<a class="headerlink" href="#python-specific-serialization" title="Permalink to this heading">¶</a></h2>
<p>It is possible to save a model in scikit-learn by using Python’s built-in
persistence model, namely <a class="reference external" href="https://docs.python.org/3/library/pickle.html">pickle</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</span> <span class="kn">import</span> <span class="n">svm</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">datasets</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">clf</span> <span class="o">=</span> <span class="n">svm</span><span class="o">.</span><span class="n">SVC</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">datasets</span><span class="o">.</span><span class="n">load_iris</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">clf</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="n">y</span><span class="p">)</span>
<span class="go">SVC()</span>

<span class="gp">&gt;&gt;&gt; </span><span class="kn">import</span> <span class="nn">pickle</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">s</span> <span class="o">=</span> <span class="n">pickle</span><span class="o">.</span><span class="n">dumps</span><span class="p">(</span><span class="n">clf</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">clf2</span> <span class="o">=</span> <span class="n">pickle</span><span class="o">.</span><span class="n">loads</span><span class="p">(</span><span class="n">s</span><span class="p">)</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">clf2</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">X</span><span class="p">[</span><span class="mi">0</span><span class="p">:</span><span class="mi">1</span><span class="p">])</span>
<span class="go">array([0])</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">y</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span>
<span class="go">0</span>
</pre></div>
</div>
<p>In the specific case of scikit-learn, it may be better to use joblib’s
replacement of pickle (<code class="docutils literal notranslate"><span class="pre">dump</span></code> &amp; <code class="docutils literal notranslate"><span class="pre">load</span></code>), which is more efficient on
objects that carry large numpy arrays internally as is often the case for
fitted scikit-learn estimators, but can only pickle to the disk and not to a
string:</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">joblib</span> <span class="kn">import</span> <span class="n">dump</span><span class="p">,</span> <span class="n">load</span>
<span class="gp">&gt;&gt;&gt; </span><span class="n">dump</span><span class="p">(</span><span class="n">clf</span><span class="p">,</span> <span class="s1">&#39;filename.joblib&#39;</span><span class="p">)</span> 
</pre></div>
</div>
<p>Later you can load back the pickled model (possibly in another Python process)
with:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">&gt;&gt;&gt; </span><span class="n">clf</span> <span class="o">=</span> <span class="n">load</span><span class="p">(</span><span class="s1">&#39;filename.joblib&#39;</span><span class="p">)</span> 
</pre></div>
</div>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p><code class="docutils literal notranslate"><span class="pre">dump</span></code> and <code class="docutils literal notranslate"><span class="pre">load</span></code> functions also accept file-like object
instead of filenames. More information on data persistence with Joblib is
available <a class="reference external" href="https://joblib.readthedocs.io/en/latest/persistence.html">here</a>.</p>
</div>
<p>When an estimator is unpickled with a scikit-learn version that is inconsistent
with the version the estimator was pickled with, a
<a class="reference internal" href="modules/generated/sklearn.exceptions.InconsistentVersionWarning.html#sklearn.exceptions.InconsistentVersionWarning" title="sklearn.exceptions.InconsistentVersionWarning"><code class="xref py py-class docutils literal notranslate"><span class="pre">InconsistentVersionWarning</span></code></a> is raised. This warning
can be caught to obtain the original version the estimator was pickled with:</p>
<blockquote>
<div><p>from sklearn.exceptions import InconsistentVersionWarning
warnings.simplefilter(“error”, InconsistentVersionWarning)</p>
<dl class="simple">
<dt>try:</dt><dd><p>est = pickle.loads(“model_from_prevision_version.pickle”)</p>
</dd>
<dt>except InconsistentVersionWarning as w:</dt><dd><p>print(w.original_sklearn_version)</p>
</dd>
</dl>
</div></blockquote>
<section id="security-maintainability-limitations">
<span id="persistence-limitations"></span><h3><span class="section-number">9.1.1. </span>Security &amp; maintainability limitations<a class="headerlink" href="#security-maintainability-limitations" title="Permalink to this heading">¶</a></h3>
<p>pickle (and joblib by extension), has some issues regarding maintainability
and security. Because of this,</p>
<ul class="simple">
<li><p>Never unpickle untrusted data as it could lead to malicious code being
executed upon loading.</p></li>
<li><p>While models saved using one version of scikit-learn might load in
other versions, this is entirely unsupported and inadvisable. It should
also be kept in mind that operations performed on such data could give
different and unexpected results.</p></li>
</ul>
<p>In order to rebuild a similar model with future versions of scikit-learn,
additional metadata should be saved along the pickled model:</p>
<ul class="simple">
<li><p>The training data, e.g. a reference to an immutable snapshot</p></li>
<li><p>The python source code used to generate the model</p></li>
<li><p>The versions of scikit-learn and its dependencies</p></li>
<li><p>The cross validation score obtained on the training data</p></li>
</ul>
<p>This should make it possible to check that the cross-validation score is in the
same range as before.</p>
<p>Aside for a few exceptions, pickled models should be portable across
architectures assuming the same versions of dependencies and Python are used.
If you encounter an estimator that is not portable please open an issue on
GitHub. Pickled models are often deployed in production using containers, like
Docker, in order to freeze the environment and dependencies.</p>
<p>If you want to know more about these issues and explore other possible
serialization methods, please refer to this
<a class="reference external" href="https://pyvideo.org/video/2566/pickles-are-for-delis-not-software">talk by Alex Gaynor</a>.</p>
</section>
<section id="a-more-secure-format-skops">
<h3><span class="section-number">9.1.2. </span>A more secure format: <code class="docutils literal notranslate"><span class="pre">skops</span></code><a class="headerlink" href="#a-more-secure-format-skops" title="Permalink to this heading">¶</a></h3>
<p><a class="reference external" href="https://skops.readthedocs.io/en/stable/">skops</a> provides a more secure
format via the <a class="reference external" href="https://skops.readthedocs.io/en/stable/modules/classes.html#module-skops.io" title="(in skops)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">skops.io</span></code></a> module. It avoids using <a class="reference external" href="https://docs.python.org/3/library/pickle.html#module-pickle" title="(in Python v3.11)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">pickle</span></code></a> and only
loads files which have types and references to functions which are trusted
either by default or by the user. The API is very similar to <code class="docutils literal notranslate"><span class="pre">pickle</span></code>, and
you can persist your models as explain in the <a class="reference external" href="https://skops.readthedocs.io/en/stable/persistence.html">docs</a> using
<a class="reference external" href="https://skops.readthedocs.io/en/stable/modules/classes.html#skops.io.dump" title="(in skops)"><code class="xref py py-func docutils literal notranslate"><span class="pre">skops.io.dump</span></code></a> and <a class="reference external" href="https://skops.readthedocs.io/en/stable/modules/classes.html#skops.io.dumps" title="(in skops)"><code class="xref py py-func docutils literal notranslate"><span class="pre">skops.io.dumps</span></code></a>:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">skops.io</span> <span class="k">as</span> <span class="nn">sio</span>
<span class="n">obj</span> <span class="o">=</span> <span class="n">sio</span><span class="o">.</span><span class="n">dumps</span><span class="p">(</span><span class="n">clf</span><span class="p">)</span>
</pre></div>
</div>
<p>And you can load them back using <a class="reference external" href="https://skops.readthedocs.io/en/stable/modules/classes.html#skops.io.load" title="(in skops)"><code class="xref py py-func docutils literal notranslate"><span class="pre">skops.io.load</span></code></a> and
<a class="reference external" href="https://skops.readthedocs.io/en/stable/modules/classes.html#skops.io.loads" title="(in skops)"><code class="xref py py-func docutils literal notranslate"><span class="pre">skops.io.loads</span></code></a>. However, you need to specify the types which are
trusted by you. You can get existing unknown types in a dumped object / file
using <a class="reference external" href="https://skops.readthedocs.io/en/stable/modules/classes.html#skops.io.get_untrusted_types" title="(in skops)"><code class="xref py py-func docutils literal notranslate"><span class="pre">skops.io.get_untrusted_types</span></code></a>, and after checking its contents,
pass it to the load function:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">unknown_types</span> <span class="o">=</span> <span class="n">sio</span><span class="o">.</span><span class="n">get_untrusted_types</span><span class="p">(</span><span class="n">obj</span><span class="p">)</span>
<span class="n">clf</span> <span class="o">=</span> <span class="n">sio</span><span class="o">.</span><span class="n">loads</span><span class="p">(</span><span class="n">obj</span><span class="p">,</span> <span class="n">trusted</span><span class="o">=</span><span class="n">unknown_types</span><span class="p">)</span>
</pre></div>
</div>
<p>If you trust the source of the file / object, you can pass <code class="docutils literal notranslate"><span class="pre">trusted=True</span></code>:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">clf</span> <span class="o">=</span> <span class="n">sio</span><span class="o">.</span><span class="n">loads</span><span class="p">(</span><span class="n">obj</span><span class="p">,</span> <span class="n">trusted</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span>
</pre></div>
</div>
<p>Please report issues and feature requests related to this format on the <a class="reference external" href="https://github.com/skops-dev/skops/issues">skops
issue tracker</a>.</p>
</section>
</section>
<section id="interoperable-formats">
<h2><span class="section-number">9.2. </span>Interoperable formats<a class="headerlink" href="#interoperable-formats" title="Permalink to this heading">¶</a></h2>
<p>For reproducibility and quality control needs, when different architectures
and environments should be taken into account, exporting the model in
<a class="reference external" href="https://onnx.ai/">Open Neural Network
Exchange</a> format or <a class="reference external" href="https://dmg.org/pmml/v4-4-1/GeneralStructure.html">Predictive Model Markup Language
(PMML)</a> format
might be a better approach than using <code class="docutils literal notranslate"><span class="pre">pickle</span></code> alone.
These are helpful where you may want to use your model for prediction in a
different environment from where the model was trained.</p>
<p>ONNX is a binary serialization of the model. It has been developed to improve
the usability of the interoperable representation of data models.
It aims to facilitate the conversion of the data
models between different machine learning frameworks, and to improve their
portability on different computing architectures. More details are available
from the <a class="reference external" href="https://onnx.ai/get-started.html">ONNX tutorial</a>.
To convert scikit-learn model to ONNX a specific tool <a class="reference external" href="http://onnx.ai/sklearn-onnx/">sklearn-onnx</a> has been developed.</p>
<p>PMML is an implementation of the <a class="reference external" href="https://en.wikipedia.org/wiki/XML">XML</a> document standard
defined to represent data models together with the data used to generate them.
Being human and machine readable,
PMML is a good option for model validation on different platforms and
long term archiving. On the other hand, as XML in general, its verbosity does
not help in production when performance is critical.
To convert scikit-learn model to PMML you can use for example <a class="reference external" href="https://github.com/jpmml/sklearn2pmml">sklearn2pmml</a> distributed under the Affero GPLv3
license.</p>
</section>
</section>


      </div>
    <div class="container">
      <footer class="sk-content-footer">
            &copy; 2007 - 2023, scikit-learn developers (BSD License).
          <a href="_sources/model_persistence.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>
$(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');
    var hide_text = 'Hide prompts and outputs';
    var show_text = 'Show prompts and outputs';

    // create and add the button to all the code blocks that contain >>>
    div.each(function(index) {
        var jthis = $(this);
        if (jthis.find('.gp').length > 0) {
            var button = $('<span class="copybutton">&gt;&gt;&gt;</span>');
            button.attr('title', hide_text);
            button.data('hidden', 'false');
            jthis.prepend(button);
        }
        // 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>');
    });

    // define the behavior of the button when it's clicked
    $('.copybutton').click(function(e){
        e.preventDefault();
        var button = $(this);
        if (button.data('hidden') === 'false') {
            // hide the code output
            button.parent().find('.go, .gp, .gt').hide();
            button.next('pre').find('.gt').nextUntil('.gp, .go').css('visibility', 'hidden');
            button.css('text-decoration', 'line-through');
            button.attr('title', show_text);
            button.data('hidden', 'true');
        } else {
            // show the code output
            button.parent().find('.go, .gp, .gt').show();
            button.next('pre').find('.gt').nextUntil('.gp, .go').css('visibility', 'visible');
            button.css('text-decoration', 'none');
            button.attr('title', hide_text);
            button.data('hidden', 'false');
        }
    });

	/*** 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>');
	});
  /*** Hide navbar when scrolling down ***/
  // Returns true when headerlink target matches hash in url
  (function() {
    hashTargetOnTop = function() {
        var hash = window.location.hash;
        if ( hash.length < 2 ) { return false; }

        var target = document.getElementById( hash.slice(1) );
        if ( target === null ) { return false; }

        var top = target.getBoundingClientRect().top;
        return (top < 2) && (top > -2);
    };

    // Hide navbar on load if hash target is on top
    var navBar = document.getElementById("navbar");
    var navBarToggler = document.getElementById("sk-navbar-toggler");
    var navBarHeightHidden = "-" + navBar.getBoundingClientRect().height + "px";
    var $window = $(window);

    hideNavBar = function() {
        navBar.style.top = navBarHeightHidden;
    };

    showNavBar = function() {
        navBar.style.top = "0";
    }

    if (hashTargetOnTop()) {
        hideNavBar()
    }

    var prevScrollpos = window.pageYOffset;
    hideOnScroll = function(lastScrollTop) {
        if (($window.width() < 768) && (navBarToggler.getAttribute("aria-expanded") === 'true')) {
            return;
        }
        if (lastScrollTop > 2 && (prevScrollpos <= lastScrollTop) || hashTargetOnTop()){
            hideNavBar()
        } else {
            showNavBar()
        }
        prevScrollpos = lastScrollTop;
    };

    /*** high performance scroll event listener***/
    var raf = window.requestAnimationFrame ||
        window.webkitRequestAnimationFrame ||
        window.mozRequestAnimationFrame ||
        window.msRequestAnimationFrame ||
        window.oRequestAnimationFrame;
    var lastScrollTop = $window.scrollTop();

    if (raf) {
        loop();
    }

    function loop() {
        var scrollTop = $window.scrollTop();
        if (lastScrollTop === scrollTop) {
            raf(loop);
            return;
        } else {
            lastScrollTop = scrollTop;
            hideOnScroll(lastScrollTop);
            raf(loop);
        }
    }
  })();
});

</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>