.. DO NOT EDIT.
.. THIS FILE WAS AUTOMATICALLY GENERATED BY SPHINX-GALLERY.
.. TO MAKE CHANGES, EDIT THE SOURCE PYTHON FILE:
.. "auto_examples/plot_classify_newsgroups.py"
.. LINE NUMBERS ARE GIVEN BELOW.

.. only:: html

    .. note::
        :class: sphx-glr-download-link-note

        :ref:`Go to the end <sphx_glr_download_auto_examples_plot_classify_newsgroups.py>`
        to download the full example code.

.. rst-class:: sphx-glr-example-title

.. _sphx_glr_auto_examples_plot_classify_newsgroups.py:


===================================================
Sparse Text Classification with Vocabulary Limits
===================================================

Shows how to pair ``TfidfVectorizer`` with ``TlmMilpClassifier`` by constraining
vocabulary dimension to prevent exponential MILP solver time, visualizing the
resulting ternary text weights.

.. GENERATED FROM PYTHON SOURCE LINES 10-62



.. image-sg:: /auto_examples/images/sphx_glr_plot_classify_newsgroups_001.png
   :alt: Exact MILP Selected Terms (Text Classification)
   :srcset: /auto_examples/images/sphx_glr_plot_classify_newsgroups_001.png
   :class: sphx-glr-single-img





.. code-block:: Python


    import matplotlib.pyplot as plt
    import numpy as np
    from sklearn.datasets import fetch_20newsgroups
    from sklearn.feature_extraction.text import TfidfVectorizer
    from sklearn.pipeline import Pipeline
    from tlmnet import TlmMilpClassifier

    # Fetch binary text dataset subset
    categories = ["alt.atheism", "sci.space"]
    newsgroups = fetch_20newsgroups(
        subset="train", categories=categories, remove=("headers", "footers", "quotes")
    )

    # Constrain vocabulary to top 25 terms to keep MILP dimension tractable
    pipeline = Pipeline(
        [
            (
                "tfidf",
                TfidfVectorizer(max_features=25, stop_words="english", sublinear_tf=True),
            ),
            ("tlm", TlmMilpClassifier(max_features=8, C=1.0)),
        ]
    )

    pipeline.fit(newsgroups.data, newsgroups.target)

    # Retrieve selected words and their discrete ternary weights
    feature_names = pipeline.named_steps["tfidf"].get_feature_names_out()
    weights = pipeline.named_steps["tlm"].coef_

    # Filter down to only the non-zero features selected by the MILP solver
    mask = weights != 0
    active_words = feature_names[mask]
    active_weights = weights[mask]

    # Sort by weight for cleaner plotting
    sort_idx = np.argsort(active_weights)
    active_words = active_words[sort_idx]
    active_weights = active_weights[sort_idx]

    # Plot the ternary weights
    plt.figure(figsize=(8, 4))
    colors = ["#d62728" if w < 0 else "#2ca02c" for w in active_weights]
    plt.barh(active_words, active_weights, color=colors, edgecolor="black")

    plt.xlabel("Ternary Weight (-1 or +1)")
    plt.title("Exact MILP Selected Terms (Text Classification)")
    plt.xticks([-1, 0, 1])
    plt.grid(axis="x", linestyle="--", alpha=0.7)
    plt.tight_layout()
    plt.show()


.. rst-class:: sphx-glr-timing

   **Total running time of the script:** (0 minutes 10.650 seconds)


.. _sphx_glr_download_auto_examples_plot_classify_newsgroups.py:

.. only:: html

  .. container:: sphx-glr-footer sphx-glr-footer-example

    .. container:: sphx-glr-download sphx-glr-download-jupyter

      :download:`Download Jupyter notebook: plot_classify_newsgroups.ipynb <plot_classify_newsgroups.ipynb>`

    .. container:: sphx-glr-download sphx-glr-download-python

      :download:`Download Python source code: plot_classify_newsgroups.py <plot_classify_newsgroups.py>`

    .. container:: sphx-glr-download sphx-glr-download-zip

      :download:`Download zipped: plot_classify_newsgroups.zip <plot_classify_newsgroups.zip>`


.. only:: html

 .. rst-class:: sphx-glr-signature

    `Gallery generated by Sphinx-Gallery <https://sphinx-gallery.github.io>`_