[ci skip] MTN Made the distinction between predictor and transformer clearer (#856) aa0fdfb

Author: ArturoAmorQ

_sources/python_scripts/02_numerical_pipeline_introduction.py

@@ -101,11 +101,18 @@
 # ![Predictor fit diagram](../figures/api_diagram-predictor.fit.svg)
 #
 # In scikit-learn an object that has a `fit` method is called an **estimator**.
+# If the estimator additionally has :
+# - a  `predict` method, it is called a **predictor**. Examples of predictors
+#   are classifiers or regressors.
+# - a `transform` method, it is called a **transformer**. Examples of
+#   transformers are scalers or encoders. We will see more about transformers in
+#   the next notebook.
+#
 # The method `fit` is composed of two elements: (i) a **learning algorithm** and
 # (ii) some **model states**. The learning algorithm takes the training data and
 # training target as input and sets the model states. These model states are
-# later used to either predict (for classifiers and regressors) or transform
-# data (for transformers).
+# later used to either predict or transform data as explained above. See the
+# glossary for more detailed definitions.
 #
 # Both the learning algorithm and the type of model states are specific to each
 # type of model.
@@ -124,8 +131,7 @@
 target_predicted = model.predict(data)
 
 # %% [markdown]
-# An estimator (an object with a `fit` method) with a `predict` method is called
-# a **predictor**. We can illustrate the prediction mechanism as follows:
+# We can illustrate the prediction mechanism as follows:
 #
 # ![Predictor predict diagram](../figures/api_diagram-predictor.predict.svg)
 #

_sources/python_scripts/02_numerical_pipeline_scaling.py

@@ -88,6 +88,7 @@
 # We show how to apply such normalization using a scikit-learn transformer
 # called `StandardScaler`. This transformer shifts and scales each feature
 # individually so that they all have a 0-mean and a unit standard deviation.
+# We recall that transformers are estimators that have a `transform` method.
 #
 # We now investigate different steps used in scikit-learn to achieve such a
 # transformation of the data.

python_scripts/02_numerical_pipeline_introduction.html

@@ -1070,11 +1070,19 @@ <h2>Fit a model and make predictions<a class="headerlink" href="#fit-a-model-and
 <p>Learning can be represented as follows:</p>
 <p><img alt="Predictor fit diagram" src="../_images/api_diagram-predictor.fit.svg" /></p>
 <p>In scikit-learn an object that has a <code class="docutils literal notranslate"><span class="pre">fit</span></code> method is called an <strong>estimator</strong>.
-The method <code class="docutils literal notranslate"><span class="pre">fit</span></code> is composed of two elements: (i) a <strong>learning algorithm</strong> and
+If the estimator additionally has :</p>
+<ul class="simple">
+<li><p>a  <code class="docutils literal notranslate"><span class="pre">predict</span></code> method, it is called a <strong>predictor</strong>. Examples of predictors
+are classifiers or regressors.</p></li>
+<li><p>a <code class="docutils literal notranslate"><span class="pre">transform</span></code> method, it is called a <strong>transformer</strong>. Examples of
+transformers are scalers or encoders. We will see more about transformers in
+the next notebook.</p></li>
+</ul>
+<p>The method <code class="docutils literal notranslate"><span class="pre">fit</span></code> is composed of two elements: (i) a <strong>learning algorithm</strong> and
 (ii) some <strong>model states</strong>. The learning algorithm takes the training data and
 training target as input and sets the model states. These model states are
-later used to either predict (for classifiers and regressors) or transform
-data (for transformers).</p>
+later used to either predict or transform data as explained above. See the
+glossary for more detailed definitions.</p>
 <p>Both the learning algorithm and the type of model states are specific to each
 type of model.</p>
 <div class="admonition note">
@@ -1091,8 +1099,7 @@ <h2>Fit a model and make predictions<a class="headerlink" href="#fit-a-model-and
 </div>
 </div>
 </div>
-<p>An estimator (an object with a <code class="docutils literal notranslate"><span class="pre">fit</span></code> method) with a <code class="docutils literal notranslate"><span class="pre">predict</span></code> method is called
-a <strong>predictor</strong>. We can illustrate the prediction mechanism as follows:</p>
+<p>We can illustrate the prediction mechanism as follows:</p>
 <p><img alt="Predictor predict diagram" src="../_images/api_diagram-predictor.predict.svg" /></p>
 <p>To predict, a model uses a <strong>prediction function</strong> that uses the input data
 together with the model states. As for the learning algorithm and the model

python_scripts/02_numerical_pipeline_scaling.html

@@ -879,7 +879,8 @@ <h2>Model fitting with preprocessing<a class="headerlink" href="#model-fitting-w
 not need such preprocessing (but would not suffer from it).</p>
 <p>We show how to apply such normalization using a scikit-learn transformer
 called <code class="docutils literal notranslate"><span class="pre">StandardScaler</span></code>. This transformer shifts and scales each feature
-individually so that they all have a 0-mean and a unit standard deviation.</p>
+individually so that they all have a 0-mean and a unit standard deviation.
+We recall that transformers are estimators that have a <code class="docutils literal notranslate"><span class="pre">transform</span></code> method.</p>
 <p>We now investigate different steps used in scikit-learn to achieve such a
 transformation of the data.</p>
 <p>First, one needs to call the method <code class="docutils literal notranslate"><span class="pre">fit</span></code> in order to learn the scaling from