.. DO NOT EDIT.
.. THIS FILE WAS AUTOMATICALLY GENERATED BY SPHINX-GALLERY.
.. TO MAKE CHANGES, EDIT THE SOURCE PYTHON FILE:
.. "auto_examples/demo_pipeline.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_demo_pipeline.py>`
        to download the full example code.

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

.. _sphx_glr_auto_examples_demo_pipeline.py:


End-to-End Pipeline Demo
========================

Demonstrates the full CoCo pipeline:
1.  Load (Tabular)
2.  Preprocess (StandardScaler)
3.  Reduce (PCA & UMAP)
4.  Report (Comparison)

Usage:
    python examples/demo_pipeline.py

.. GENERATED FROM PYTHON SOURCE LINES 15-132







.. code-block:: Python


    import logging
    from pathlib import Path

    import numpy as np
    import pandas as pd

    from coco_pipe.dim_reduction import DimReduction
    from coco_pipe.io.dataset import TabularDataset
    from coco_pipe.report import from_reductions

    # Configure logging
    logging.basicConfig(level=logging.INFO)
    logger = logging.getLogger(__name__)


    def _sample_metadata(container) -> dict[str, np.ndarray]:
        """Extract observation-aligned metadata columns from a DataContainer."""
        n_samples = np.asarray(container.X).shape[0]
        metadata = {}
        for key, value in (getattr(container, "coords", {}) or {}).items():
            value = np.asarray(value)
            if key == "obs":
                continue
            if value.ndim >= 1 and value.shape[0] == n_samples:
                metadata[key] = value
        return metadata


    def create_synthetic_csv(path: Path):
        """Create a dummy CSV file for the demo."""
        logger.info(f"Creating synthetic dataset at {path}...")

        # 3 Clusters
        c1 = np.random.randn(50, 10) + 5
        c2 = np.random.randn(50, 10) - 5
        c3 = np.random.randn(100, 10)

        X = np.vstack([c1, c2, c3])
        labels = ["A"] * 50 + ["B"] * 50 + ["C"] * 100

        df = pd.DataFrame(X, columns=[f"feat_{i}" for i in range(10)])
        df["label"] = labels
        # Add extra metadata for interactive labeling demo
        df["center"] = np.random.choice(["Site A", "Site B"], size=len(df))
        df["batch"] = np.random.choice(["Batch 1", "Batch 2"], size=len(df))

        df.to_csv(path, index=False)


    def main():
        # Setup
        data_path = Path("examples/outputs/dummy_data.csv")
        data_path.parent.mkdir(parents=True, exist_ok=True)
        create_synthetic_csv(data_path)

        # 1. Load Data
        logger.info("1. Loading Data...")
        ds = TabularDataset(
            data_path, target_col="label", sep=",", meta_columns=["center", "batch"]
        )
        container = ds.load()

        # 2. Preprocessing (simulate via manual scaling)
        logger.info("2. Preprocessing...")
        # Ensure X is numeric
        if isinstance(container.X, pd.DataFrame):
            container.X = container.X.select_dtypes(include=[np.number])
        elif hasattr(container.X, "dtype") and container.X.dtype == object:
            container.X = (
                pd.DataFrame(container.X).select_dtypes(include=[np.number]).values
            )

        container.X = (container.X - container.X.mean(axis=0)) / container.X.std(axis=0)

        # 3. Dimensionality Reduction
        logger.info("3. Running Reductions...")

        # PCA
        pca = DimReduction(method="PCA", n_components=2)
        pca_emb = pca.fit_transform(container.X)

        # UMAP
        umap = DimReduction(method="UMAP", n_components=2, n_neighbors=15)
        umap_emb = umap.fit_transform(container.X)

        # 4. Generate Comparative Report
        logger.info("4. Generating Report...")

        # Calculate dummy metrics for demonstration
        metrics_data = {
            "Trustworthiness": [0.95, 0.88],
            "Continuity": [0.90, 0.92],
            "Shepard Goodness": [0.85, 0.91],
        }
        metrics_df = pd.DataFrame(metrics_data, index=["PCA", "UMAP"])

        report = from_reductions(
            reductions=[pca, umap],
            container=container,
            embeddings=[pca_emb, umap_emb],
            labels=container.y,
            metadata=_sample_metadata(container),
            title="Pipeline Demo: PCA vs UMAP",
            config={"pipeline": "Full Demo", "scaling": "StandardScaler"},
        )

        # Add comparison section
        report.add_comparison(metrics_df, name="Method Comparison")

        output_path = Path("examples/outputs/demo_pipeline.html")
        report.save(output_path)
        logger.info(f"Report saved to {output_path}")


    if __name__ == "__main__":
        main()


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

   **Total running time of the script:** (0 minutes 0.416 seconds)


.. _sphx_glr_download_auto_examples_demo_pipeline.py:

.. only:: html

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

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

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

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

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

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

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


.. only:: html

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

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