bigframes 2.49.0__tar.gz → 2.51.0__tar.gz

{bigframes-2.49.0/bigframes.egg-info → bigframes-2.51.0}/PKG-INFO

@@ -1,15 +1,15 @@
 Metadata-Version: 2.4
 Name: bigframes
-Version: 2.49.0
+Version: 2.51.0
 Summary: BigQuery DataFrames -- scalable analytics and machine learning with BigQuery
 Home-page: https://dataframes.bigquery.dev
-Download-URL: https://github.com/googleapis/google-cloud-python/tree/main/packages/bigframes/releases
+Download-URL: https://github.com/google/bigframes/releases
 Author: Google LLC
 Author-email: bigframes-feedback@google.com
 License: Apache 2.0
-Project-URL: Source, https://github.com/googleapis/google-cloud-python/tree/main/packages/bigframes
+Project-URL: Source, https://github.com/google/bigframes
 Project-URL: Changelog, https://dataframes.bigquery.dev/changelog.html
-Project-URL: Issues, https://github.com/googleapis/google-cloud-python/tree/main/packages/bigframes/issues
+Project-URL: Issues, https://github.com/google/bigframes/issues
 Platform: Posix; MacOS X; Windows
 Classifier: Development Status :: 5 - Production/Stable
 Classifier: Intended Audience :: Developers
@@ -31,7 +31,7 @@ Requires-Dist: fsspec>=2023.3.0
 Requires-Dist: gcsfs!=2025.5.0,!=2026.2.0,!=2026.3.0,>=2023.3.0
 Requires-Dist: geopandas>=0.12.2
 Requires-Dist: google-auth[pyopenssl]<3.0,>=2.15.0
-Requires-Dist: google-cloud-bigquery[bqstorage,pandas]>=3.36.0
+Requires-Dist: google-cloud-bigquery[bqstorage,pandas]>=3.42.0
 Requires-Dist: google-cloud-bigquery-storage<3.0.0,>=2.30.0
 Requires-Dist: google-cloud-functions>=1.20.2
 Requires-Dist: google-cloud-bigquery-connection>=1.18.2
@@ -118,12 +118,12 @@ for many use cases, including:
 * `bigframes.pandas <https://dataframes.bigquery.dev/reference/api/bigframes.pandas.html>`_
   is a pandas API for analytics. Many workloads can be
   migrated from pandas to bigframes by just changing a few imports.
+* `bigframes.bigquery.ai <https://dataframes.bigquery.dev/reference/api/bigframes.bigquery.ai.html>`_
+  provides powerful AI functions, text generation, vector embeddings, and semantic search powered by Gemini.
 * `bigframes.ml <https://dataframes.bigquery.dev/reference/index.html#ml-apis>`_
   is a scikit-learn-like API for ML.
-* `bigframes.bigquery.ai <https://dataframes.bigquery.dev/reference/api/bigframes.bigquery.ai.html>`_
-  are a collection of powerful AI methods, powered by Gemini.
 
-BigQuery DataFrames is an `open-source package <https://github.com/googleapis/google-cloud-python>`_.
+BigQuery DataFrames is an `open-source package <https://github.com/google/bigframes>`_.
 
 .. |GA| image:: https://img.shields.io/badge/support-GA-gold.svg
    :target: https://github.com/googleapis/google-cloud-python/blob/main/README.rst#general-availability
@@ -132,21 +132,42 @@ BigQuery DataFrames is an `open-source package <https://github.com/googleapis/go
 .. |versions| image:: https://img.shields.io/pypi/pyversions/bigframes.svg
    :target: https://pypi.org/project/bigframes/
 
+Why Use BigQuery DataFrames?
+----------------------------
+
+BigQuery DataFrames eliminates the "data movement bottleneck." Instead of downloading large datasets to a local environment, BigFrames translates your Python code into optimized SQL executed across BigQuery's distributed compute engine.
+
+* **Petabyte-Scale Scalability:** Effortlessly process datasets exceeding local memory constraints without provisioning or managing Spark or Ray clusters.
+* **Familiar Python Ecosystem:** Use the pandas operations you already know (``read_gbq``, ``groupby``, ``merge``, ``pivot_table``) directly on BigQuery tables.
+* **Generative AI & Multimodality:** Seamlessly leverage Gemini models, AI functions, multimodal processing (audio, video, PDF), vector embeddings, and vector search with ``bigframes.bigquery.ai``.
+* **Optimized Execution:** Take advantage of BigFrames partial ordering mode (``bpd.options.bigquery.ordering_mode = "partial"``) for considerable performance gains on analytical workloads.
+* **Enterprise Security:** Keep your data securely governed within the BigQuery security perimeter.
+* **Hybrid Flexibility:** Seamlessly transition between distributed BigQuery processing and local pandas analysis with ``to_pandas()`` and ``read_pandas()``.
+
 Getting started with BigQuery DataFrames
 ----------------------------------------
 
-The easiest way to get started is to try the
-`BigFrames quickstart <https://cloud.google.com/bigquery/docs/dataframes-quickstart>`_
-in a `notebook in BigQuery Studio <https://cloud.google.com/bigquery/docs/notebooks-introduction>`_.
+Notebooks
+~~~~~~~~~
+
+The easiest way to get started is to open a sample notebook in
+`Colab <https://colab.research.google.com/github/google/bigframes/blob/main/notebooks/getting_started/bq_dataframes_template.ipynb>`_
+or
+`BigQuery Studio <https://console.cloud.google.com/bigquery/import?url=https://github.com/google/bigframes/blob/main/notebooks/getting_started/bq_dataframes_template.ipynb>`_.
+More details can be found in the `BigFrames quickstart <https://cloud.google.com/bigquery/docs/dataframes-quickstart>`_
+and in the `introduction to notebooks <https://cloud.google.com/bigquery/docs/notebooks-introduction>`_.
+
+Local environment
+~~~~~~~~~~~~~~~~~
 
-To use BigFrames in your local development environment,
+To use BigFrames in your local development environment:
 
 1. Run ``pip install --upgrade bigframes`` to install the latest version.
 
-2. Setup `Application default credentials <https://cloud.google.com/docs/authentication/set-up-adc-local-dev-environment>`_
-   for your local development environment enviroment.
+2. Set up `Application Default Credentials <https://cloud.google.com/docs/authentication/set-up-adc-local-dev-environment>`_
+   for your local development environment.
 
-3. Create a `GCP project with the BigQuery API enabled <https://cloud.google.com/bigquery/docs/sandbox>`_.
+3. Create a `Google Cloud project with the BigQuery API enabled <https://cloud.google.com/bigquery/docs/sandbox>`_.
 
 4. Use the ``bigframes`` package to query data.
 
@@ -171,15 +192,15 @@ Documentation
 To learn more about BigQuery DataFrames, visit these pages
 
 * `Introduction to BigQuery DataFrames (BigFrames) <https://cloud.google.com/bigquery/docs/bigquery-dataframes-introduction>`_
-* `Sample notebooks <https://github.com/googleapis/google-cloud-python/tree/main/notebooks>`_
+* `Sample notebooks <https://github.com/google/bigframes/tree/main/notebooks>`_
 * `API reference <https://dataframes.bigquery.dev/>`_
-* `Source code (GitHub) <https://github.com/googleapis/google-cloud-python>`_
+* `Source code (GitHub) <https://github.com/google/bigframes>`_
 
 License
 -------
 
 BigQuery DataFrames is distributed with the `Apache-2.0 license
-<https://github.com/googleapis/google-cloud-python/blob/main/LICENSE>`_.
+<https://github.com/google/bigframes/blob/main/LICENSE>`_.
 
 It also contains code derived from the following third-party packages:
 
@@ -191,7 +212,7 @@ It also contains code derived from the following third-party packages:
 * `SQLGlot <https://sqlglot.com/sqlglot.html>`_
 
 For details, see the `third_party
-<https://github.com/googleapis/google-cloud-python/tree/main/third_party/bigframes_vendored>`_
+<https://github.com/google/bigframes/tree/main/third_party/bigframes_vendored>`_
 directory.
 
 

{bigframes-2.49.0 → bigframes-2.51.0}/README.rst

@@ -11,12 +11,12 @@ for many use cases, including:
 * `bigframes.pandas <https://dataframes.bigquery.dev/reference/api/bigframes.pandas.html>`_
   is a pandas API for analytics. Many workloads can be
   migrated from pandas to bigframes by just changing a few imports.
+* `bigframes.bigquery.ai <https://dataframes.bigquery.dev/reference/api/bigframes.bigquery.ai.html>`_
+  provides powerful AI functions, text generation, vector embeddings, and semantic search powered by Gemini.
 * `bigframes.ml <https://dataframes.bigquery.dev/reference/index.html#ml-apis>`_
   is a scikit-learn-like API for ML.
-* `bigframes.bigquery.ai <https://dataframes.bigquery.dev/reference/api/bigframes.bigquery.ai.html>`_
-  are a collection of powerful AI methods, powered by Gemini.
 
-BigQuery DataFrames is an `open-source package <https://github.com/googleapis/google-cloud-python>`_.
+BigQuery DataFrames is an `open-source package <https://github.com/google/bigframes>`_.
 
 .. |GA| image:: https://img.shields.io/badge/support-GA-gold.svg
    :target: https://github.com/googleapis/google-cloud-python/blob/main/README.rst#general-availability
@@ -25,21 +25,42 @@ BigQuery DataFrames is an `open-source package <https://github.com/googleapis/go
 .. |versions| image:: https://img.shields.io/pypi/pyversions/bigframes.svg
    :target: https://pypi.org/project/bigframes/
 
+Why Use BigQuery DataFrames?
+----------------------------
+
+BigQuery DataFrames eliminates the "data movement bottleneck." Instead of downloading large datasets to a local environment, BigFrames translates your Python code into optimized SQL executed across BigQuery's distributed compute engine.
+
+* **Petabyte-Scale Scalability:** Effortlessly process datasets exceeding local memory constraints without provisioning or managing Spark or Ray clusters.
+* **Familiar Python Ecosystem:** Use the pandas operations you already know (``read_gbq``, ``groupby``, ``merge``, ``pivot_table``) directly on BigQuery tables.
+* **Generative AI & Multimodality:** Seamlessly leverage Gemini models, AI functions, multimodal processing (audio, video, PDF), vector embeddings, and vector search with ``bigframes.bigquery.ai``.
+* **Optimized Execution:** Take advantage of BigFrames partial ordering mode (``bpd.options.bigquery.ordering_mode = "partial"``) for considerable performance gains on analytical workloads.
+* **Enterprise Security:** Keep your data securely governed within the BigQuery security perimeter.
+* **Hybrid Flexibility:** Seamlessly transition between distributed BigQuery processing and local pandas analysis with ``to_pandas()`` and ``read_pandas()``.
+
 Getting started with BigQuery DataFrames
 ----------------------------------------
 
-The easiest way to get started is to try the
-`BigFrames quickstart <https://cloud.google.com/bigquery/docs/dataframes-quickstart>`_
-in a `notebook in BigQuery Studio <https://cloud.google.com/bigquery/docs/notebooks-introduction>`_.
+Notebooks
+~~~~~~~~~
+
+The easiest way to get started is to open a sample notebook in
+`Colab <https://colab.research.google.com/github/google/bigframes/blob/main/notebooks/getting_started/bq_dataframes_template.ipynb>`_
+or
+`BigQuery Studio <https://console.cloud.google.com/bigquery/import?url=https://github.com/google/bigframes/blob/main/notebooks/getting_started/bq_dataframes_template.ipynb>`_.
+More details can be found in the `BigFrames quickstart <https://cloud.google.com/bigquery/docs/dataframes-quickstart>`_
+and in the `introduction to notebooks <https://cloud.google.com/bigquery/docs/notebooks-introduction>`_.
+
+Local environment
+~~~~~~~~~~~~~~~~~
 
-To use BigFrames in your local development environment,
+To use BigFrames in your local development environment:
 
 1. Run ``pip install --upgrade bigframes`` to install the latest version.
 
-2. Setup `Application default credentials <https://cloud.google.com/docs/authentication/set-up-adc-local-dev-environment>`_
-   for your local development environment enviroment.
+2. Set up `Application Default Credentials <https://cloud.google.com/docs/authentication/set-up-adc-local-dev-environment>`_
+   for your local development environment.
 
-3. Create a `GCP project with the BigQuery API enabled <https://cloud.google.com/bigquery/docs/sandbox>`_.
+3. Create a `Google Cloud project with the BigQuery API enabled <https://cloud.google.com/bigquery/docs/sandbox>`_.
 
 4. Use the ``bigframes`` package to query data.
 
@@ -64,15 +85,15 @@ Documentation
 To learn more about BigQuery DataFrames, visit these pages
 
 * `Introduction to BigQuery DataFrames (BigFrames) <https://cloud.google.com/bigquery/docs/bigquery-dataframes-introduction>`_
-* `Sample notebooks <https://github.com/googleapis/google-cloud-python/tree/main/notebooks>`_
+* `Sample notebooks <https://github.com/google/bigframes/tree/main/notebooks>`_
 * `API reference <https://dataframes.bigquery.dev/>`_
-* `Source code (GitHub) <https://github.com/googleapis/google-cloud-python>`_
+* `Source code (GitHub) <https://github.com/google/bigframes>`_
 
 License
 -------
 
 BigQuery DataFrames is distributed with the `Apache-2.0 license
-<https://github.com/googleapis/google-cloud-python/blob/main/LICENSE>`_.
+<https://github.com/google/bigframes/blob/main/LICENSE>`_.
 
 It also contains code derived from the following third-party packages:
 
@@ -84,7 +105,7 @@ It also contains code derived from the following third-party packages:
 * `SQLGlot <https://sqlglot.com/sqlglot.html>`_
 
 For details, see the `third_party
-<https://github.com/googleapis/google-cloud-python/tree/main/third_party/bigframes_vendored>`_
+<https://github.com/google/bigframes/tree/main/third_party/bigframes_vendored>`_
 directory.
 
 

{bigframes-2.49.0 → bigframes-2.51.0}/bigframes/bigquery/_operations/ai.py

@@ -19,7 +19,18 @@ https://cloud.google.com/bigquery/docs/reference/standard-sql/bigqueryml-syntax-
 from __future__ import annotations
 
 import json
-from typing import Any, Dict, Iterable, List, Literal, Mapping, Optional, Tuple, Union
+from typing import (
+    Any,
+    Dict,
+    Iterable,
+    List,
+    Literal,
+    Mapping,
+    Optional,
+    Tuple,
+    Union,
+    overload,
+)
 
 import pandas as pd
 
@@ -1152,7 +1163,7 @@ def forecast(
         if column not in df.columns:
             raise ValueError(f"Column `{column}` not found")
 
-    options: dict[str, Union[int, float, str, Iterable[str]]] = {
+    options: dict[str, int | float | str | Iterable[str]] = {
         "data_col": data_col,
         "timestamp_col": timestamp_col,
         "model": model,
@@ -1168,6 +1179,267 @@ def forecast(
     return ml_core.BaseBqml(df._session).ai_forecast(input_data=df, options=options)
 
 
+@log_adapter.method_logger(custom_base_name="bigquery_ai")
+def predict(
+    training_df: dataframe.DataFrame | pd.DataFrame,
+    prediction_df: dataframe.DataFrame | pd.DataFrame,
+    *,
+    label_col: str = "label",
+) -> dataframe.DataFrame:
+    """
+    Uses TabFm, a pre-trained foundation model for tabular data, to perform regression
+    and classification tasks on structured data.
+
+    **Examples:**
+        >>> df = bpd.read_gbq("bigquery-public-data.ml_datasets.penguins")
+        >>> df = df[df['body_mass_g'] > 0]
+
+        >>> size = len(df)
+        >>> training_size = int(size * 0.8)
+        >>> training_df = df.head(training_size)
+        >>> prediction_df = df.tail(size - training_size).dropna(subset=['body_mass_g'])
+
+        >>> result = bbq.ai.predict(training_df,prediction_df,label_col='body_mass_g')
+        >>> type(result) # doctest: +ELLIPSIS
+        <class 'pandas...DataFrame'>
+
+    Args:
+        training_df (DataFrame):
+            The dataframe that contains the training data. It could be either a
+            BigFrames Dataframe or a pandas DataFrame. If it's a pandas
+            DataFrame, the global BigQuery session will be used to load the
+            data. The table or query result must contain a column named label or
+            the column that you specify in the 'label_col' argument. Every other
+            column is considered a feature column. The feature and label columns
+            must be one of the following types: string, bool, int, float, or
+            decimal
+        prediction_df (DataFrame):
+            The table or query that contains the data to run prediction on. It
+            could be either a BigFrames Dataframe or a pandas DataFrame. If it's
+            a pandas DataFrame, the global BigQuery session will be used to load
+            the data. The table or query result must contain all of the feature
+            columns in the training data and can optionally contain additional
+            columns.
+        label_col (str, default 'label'):
+            A string value that specifies the name of the label column in the
+            training data.
+    """
+    session, training_df, prediction_df = _to_bf_dataframes(training_df, prediction_df)
+
+    return ml_core.BaseBqml(session).ai_predict(
+        training_df, prediction_df, options={"label_col": label_col}
+    )
+
+
+@overload
+def evaluate(
+    df1: dataframe.DataFrame | pd.DataFrame,
+    df2: dataframe.DataFrame | pd.DataFrame,
+    *,
+    data_col: str,
+    timestamp_col: str,
+    model: str = "TimesFM 2.5",
+    id_cols: Iterable[str] | None = None,
+    horizon: int = 1024,
+    context_window: int | None = None,
+) -> dataframe.DataFrame: ...
+
+
+@overload
+def evaluate(
+    df1: dataframe.DataFrame | pd.DataFrame,
+    df2: dataframe.DataFrame | pd.DataFrame,
+    *,
+    label_col: str,
+) -> dataframe.DataFrame: ...
+
+
+@log_adapter.method_logger(custom_base_name="bigquery_ai")
+def evaluate(
+    df1: dataframe.DataFrame | pd.DataFrame,
+    df2: dataframe.DataFrame | pd.DataFrame,
+    *,
+    data_col: str | None = None,
+    timestamp_col: str | None = None,
+    model: str = "TimesFM 2.5",
+    id_cols: Iterable[str] | None = None,
+    horizon: int = 1024,
+    context_window: int | None = None,
+    label_col: str | None = None,
+) -> dataframe.DataFrame:
+    """
+    Evaluates TimesFM forecasted data against a reference time series based on
+    historical data, or TabFM predicted data against ground truth data.
+
+    **Examples:**
+
+        Evaluate TimesFM forecasted data:
+
+        >>> import pandas as pd
+        >>> import bigframes.pandas as bpd
+        >>> import bigframes.bigquery as bbq
+        >>> history_df = bpd.DataFrame({
+        ...     "value": [1.0, 2.0, 3.0, 4.0],
+        ...     "time": pd.to_datetime(["2020-01-01", "2020-01-02", "2020-01-03", "2020-01-04"]),
+        ... })
+        >>> actual_df = bpd.DataFrame({
+        ...     "value": [5.0, 6.0],
+        ...     "time": pd.to_datetime(["2020-01-05", "2020-01-06"]),
+        ... })
+        >>> result = bbq.ai.evaluate(
+        ...     history_df,
+        ...     actual_df,
+        ...     data_col="value",
+        ...     timestamp_col="time",
+        ... )
+        >>> type(result)
+        <class 'bigframes.dataframe.DataFrame'>
+
+        Evaluate TabFM predicted data:
+
+        >>> df = bpd.read_gbq("bigquery-public-data.ml_datasets.penguins")
+        >>> df = df[df["body_mass_g"] > 0]
+        >>> size = len(df)
+        >>> training_size = int(size * 0.8)
+        >>> training_df = df.head(training_size)
+        >>> prediction_df = df.tail(size - training_size).dropna(subset=["body_mass_g"])
+        >>> result = bbq.ai.evaluate(
+        ...     training_df,
+        ...     prediction_df,
+        ...     label_col="body_mass_g",
+        ... )
+        >>> type(result)
+        <class 'bigframes.dataframe.DataFrame'>
+
+    Args:
+        df1 (DataFrame):
+            For TimesFM evaluation, the dataframe containing historical time
+            series data used to generate a forecast. For TabFM evaluation, the
+            dataframe containing training data (must contain ``label_col``;
+            every other column is considered a feature column of type
+            ``STRING``, ``BOOL``, ``INT64``, ``FLOAT64``, ``NUMERIC``, or
+            ``BIGNUMERIC``). Can be either a BigFrames DataFrame or a pandas
+            DataFrame. If it's a pandas DataFrame, the global BigQuery session
+            will be used to load the data.
+        df2 (DataFrame):
+            For TimesFM evaluation, the dataframe containing actual time series
+            data to evaluate against forecasted values. For TabFM evaluation,
+            the dataframe containing the data to run prediction on and evaluate
+            against ``label_col`` (must contain all feature columns in the
+            training data and ``label_col``, and can optionally contain
+            additional columns). Can be either a BigFrames DataFrame or a pandas
+            DataFrame. If it's a pandas DataFrame, the global BigQuery session
+            will be used to load the data.
+        data_col (str, optional):
+            A ``str`` value that specifies the name of the time series data
+            column (required for TimesFM evaluation). The data column must use
+            one of the following data types: ``INT64``, ``NUMERIC``,
+            ``BIGNUMERIC``, or ``FLOAT64``.
+        timestamp_col (str, optional):
+            A ``str`` value that specifies the name of the timestamp column
+            (required for TimesFM evaluation). The timestamp column must use one
+            of the following data types: ``TIMESTAMP``, ``DATE``, or ``DATETIME``.
+        model (str, default "TimesFM 2.5"):
+            A ``str`` value that specifies the name of the model to use for
+            TimesFM evaluation. Supported models include ``"TimesFM 2.0"`` and
+            ``"TimesFM 2.5"``. The default value is ``"TimesFM 2.5"``, which is
+            recommended for all new evaluation tasks.
+        id_cols (Iterable[str], optional):
+            An iterable of ``str`` values that specifies the names of one or
+            more ID columns for TimesFM evaluation. Each unique combination of
+            IDs identifies a unique time series to evaluate. Specify one or more
+            values for this argument in order to evaluate multiple time series
+            using a single query. The columns that you specify must use one of
+            the following data types: ``STRING`` or ``INT64``.
+        horizon (int, default 1024):
+            An ``int`` value that specifies the number of forecasted time points
+            to evaluate for TimesFM evaluation. The default value is ``1024``.
+            The valid input range is ``[1, 10,000]``.
+        context_window (int, optional):
+            An ``int`` value that specifies the context window length used by
+            BigQuery ML's built-in TimesFM model. The context window length
+            determines how many of the most recent data points from the input
+            time series are used by the model. If you don't specify a value, the
+            ``AI.EVALUATE`` function automatically chooses the smallest possible
+            context window length to use that is still large enough to cover the
+            number of time series data points in your input data.
+        label_col (str, optional):
+            A ``str`` value that specifies the name of the label column in the
+            training data (required for TabFM evaluation). If the column is of
+            type ``STRING`` or ``BOOL``, then classification is evaluated. If
+            the column is of type ``INT64``, ``FLOAT64``, ``NUMERIC``, or
+            ``BIGNUMERIC``, then regression is evaluated.
+
+    Returns:
+        bigframes.pandas.DataFrame:
+            A DataFrame containing the evaluation metrics from BigQuery
+            ``AI.EVALUATE``. See
+            https://cloud.google.com/bigquery/docs/reference/standard-sql/bigqueryml-syntax-ai-evaluate
+
+    Raises:
+        ValueError: When invalid combinations of parameters are specified or
+            when any specified column does not exist in the input dataframes.
+    """
+    # 1. TabFM mode (tabular prediction evaluation)
+    if label_col is not None:
+        if data_col is not None or timestamp_col is not None:
+            raise ValueError(
+                "Cannot specify both `label_col` and `data_col`/`timestamp_col`."
+            )
+        required_cols = [label_col]
+        options: dict[str, int | float | str | Iterable[str]] = {
+            "label_col": label_col,
+        }
+    # 2. TimesFM mode (time series forecast evaluation)
+    elif data_col is not None and timestamp_col is not None:
+        required_cols = [timestamp_col, data_col, *(id_cols or ())]
+        options = {
+            "data_col": data_col,
+            "timestamp_col": timestamp_col,
+            "model": model,
+            "horizon": horizon,
+        }
+        if id_cols:
+            options["id_cols"] = id_cols
+        if context_window is not None:
+            options["context_window"] = context_window
+    else:
+        raise ValueError(
+            "Must specify either (`data_col` and `timestamp_col`) for time series "
+            "evaluation or `label_col` for tabular evaluation."
+        )
+
+    for col in required_cols:
+        if col not in df1.columns or col not in df2.columns:
+            raise ValueError(f"Column `{col}` not found")
+
+    session, df1, df2 = _to_bf_dataframes(df1, df2)
+
+    return ml_core.BaseBqml(session).ai_evaluate(df1, df2, options=options)
+
+
+def _to_bf_dataframes(
+    df1: dataframe.DataFrame | pd.DataFrame,
+    df2: dataframe.DataFrame | pd.DataFrame,
+) -> Tuple[session.Session, dataframe.DataFrame, dataframe.DataFrame]:
+    """Find a unifying session for two DataFrames and convert pandas DataFrames to BigFrames DataFrames."""
+    unifying_session = None
+    for df in [df1, df2]:
+        if isinstance(df, dataframe.DataFrame):
+            unifying_session = df._session
+            break
+
+    if unifying_session is None:
+        unifying_session = bpd.get_global_session()
+
+    if isinstance(df1, pd.DataFrame):
+        df1 = unifying_session.read_pandas(df1)
+    if isinstance(df2, pd.DataFrame):
+        df2 = unifying_session.read_pandas(df2)
+
+    return unifying_session, df1, df2
+
+
 def _separate_context_and_series(
     prompt: PROMPT_TYPE,
 ) -> Tuple[List[str | None], List[series.Series]]: