julearn 0.3.5.dev140__tar.gz → 0.3.6.dev15__tar.gz

{julearn-0.3.5.dev140 → julearn-0.3.6.dev15}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: julearn
-Version: 0.3.5.dev140
+Version: 0.3.6.dev15
 Summary: Juelich Machine Learning Library
 Author-email: Fede Raimondo <f.raimondo@fz-juelich.de>, Sami Hamdan <s.hamdan@fz-juelich.de>
 Maintainer-email: Sami Hamdan <s.hamdan@fz-juelich.de>

julearn-0.3.6.dev15/docs/changes/newsfragments/301.misc

@@ -0,0 +1 @@
+Delay log message formatting by `Fede Raimondo`_

julearn-0.3.6.dev15/docs/changes/newsfragments/303.feature

@@ -0,0 +1 @@
+Allow setting of ``context_func`` for ``joblib-htcondor`` by `Fede Raimondo`_

{julearn-0.3.5.dev140 → julearn-0.3.6.dev15}/docs/configuration.rst

@@ -22,15 +22,18 @@ Here you can find the comprehensive list of flags that can be set:
    :header-rows: 1
 
    * - Flag
+     - Value
      - Description
      - Potential problem(s)
    * - ``disable_x_check``
+     - ``False``
      - | Disable checking for unmatched column names in ``X``.
        | If set to ``True``, any element in ``X`` that is not present in the
        | dataframe will not result in an error.
      - | The user might think that a certain feature is used in the model when
        | it is not.
    * - ``disable_xtypes_check``
+     - ``False``
      - | Disable checking for missing/present ``X_types`` in the ``X`` parameter
        | of the :func:`.run_cross_validation` method.
        | If set to ``True``, the ``X_types`` parameter will not be checked for
@@ -40,14 +43,17 @@ Here you can find the comprehensive list of flags that can be set:
      - | The user might think that a certain feature is considered in the model
        | when it is not.
    * - ``disable_x_verbose``
+     - ``False``
      - | Disable printing the list of expanded column names in ``X``.
        | If set to ``True``, the list of column names will not be printed.
      - The user will not see the expanded column names in ``X``.
    * - ``disable_xtypes_verbose``
+     - ``False``
      - | Disable printing the list of expanded column names in ``X_types``.
        | If set to ``True``, the list of types of X will not be printed.
      - The user will not see the expanded ``X_types`` column names.
    * - ``enable_parallel_column_transformers``
+     - ``False``
      - | This flag enables parallel execution of column transformers by
        | reverting to the default behaviour of scikit-learn
        | (instead of using ``n_jobs=1``)
@@ -55,6 +61,13 @@ Here you can find the comprehensive list of flags that can be set:
      - | Column transformers will be applied in parallel, using more resources.
        | than expected.
    * - ``enable_auto_escape_parenthesis``
+     - ``True``
      - | This flag enables escaping parenthesis in column names.
      - | If column names include parenthesis, then they will be treated as regular
        | expression metacharacters and not regular characters.
+   * - ``max_x_logs``
+     - ``100``
+     - | This flag sets the maximum number of columns in ``X`` for which the
+       | column names will be printed.
+     - | If ``X`` has more columns than this value, the column names will
+       | not be printed, and a the number of columns will be displayed instead.

julearn-0.3.6.dev15/julearn/_version.py

@@ -0,0 +1,24 @@
+# file generated by vcs-versioning
+# don't change, don't track in version control
+from __future__ import annotations
+
+__all__ = [
+    "__version__",
+    "__version_tuple__",
+    "version",
+    "version_tuple",
+    "__commit_id__",
+    "commit_id",
+]
+
+version: str
+__version__: str
+__version_tuple__: tuple[int | str, ...]
+version_tuple: tuple[int | str, ...]
+commit_id: str | None
+__commit_id__: str | None
+
+__version__ = version = '0.3.6.dev15'
+__version_tuple__ = version_tuple = (0, 3, 6, 'dev15')
+
+__commit_id__ = commit_id = 'g4baf848be'

{julearn-0.3.5.dev140 → julearn-0.3.6.dev15}/julearn/api.py

@@ -10,6 +10,7 @@ from typing import Optional, Union
 import numpy as np
 import pandas as pd
 import sklearn
+from joblib.parallel import get_active_backend
 from sklearn.base import BaseEstimator
 from sklearn.model_selection import (
     cross_validate,
@@ -17,6 +18,7 @@ from sklearn.model_selection import (
 from sklearn.model_selection._search import BaseSearchCV
 from sklearn.pipeline import Pipeline
 
+from .config import _joblib_htcondor_context_func
 from .inspect import Inspector
 from .model_selection.utils import check_cv
 from .pipeline import PipelineCreator
@@ -24,6 +26,7 @@ from .pipeline.merger import merge_pipelines
 from .prepare import check_consistency, prepare_input_data
 from .scoring import check_scoring
 from .utils import _compute_cvmdsum, logger, raise_error
+from .utils.logging import DelayedFmtMessage as __
 from .utils.typing import CVLike
 
 
@@ -185,7 +188,7 @@ def _validate_api_params(  # noqa: C901
     if seed is not None:
         # If a seed is passed, use it, otherwise do not do anything. User
         # might have set the seed outside of the library
-        logger.info(f"Setting random seed to {seed}")
+        logger.info(__("Setting random seed to {seed}", seed=seed))
         np.random.seed(seed)
 
     # Interpret the input data and prepare it to be used with the library
@@ -341,18 +344,32 @@ def _validate_api_params(  # noqa: C901
 
     # Log some information
     logger.info("= Data Information =")
-    logger.info(f"\tProblem type: {problem_type}")
-    logger.info(f"\tNumber of samples: {len(df_X)}")
-    logger.info(f"\tNumber of features: {len(df_X.columns)}")
+    logger.info(
+        __("\tProblem type: {problem_type}", problem_type=problem_type)
+    )
+    logger.info(__("\tNumber of samples: {n_samples}", n_samples=len(df_X)))
+    logger.info(
+        __("\tNumber of features: {n_features}", n_features=len(df_X.columns))
+    )
     logger.info("====================")
     logger.info("")
 
     if problem_type == "classification":
-        logger.info(f"\tNumber of classes: {len(np.unique(df_y))}")
-        logger.info(f"\tTarget type: {df_y.dtype}")
-        logger.info(f"\tClass distributions: {df_y.value_counts()}")
+        logger.info(
+            __(
+                "\tNumber of classes: {n_classes}",
+                n_classes=len(np.unique(df_y)),
+            )
+        )
+        logger.info(__("\tTarget type: {target_type}", target_type=df_y.dtype))
+        logger.info(
+            __(
+                "\tClass distributions: {class_distributions}",
+                class_distributions=df_y.value_counts(),
+            )
+        )
     elif problem_type == "regression":
-        logger.info(f"\tTarget type: {df_y.dtype}")
+        logger.info(__("\tTarget type: {target_type}", target_type=df_y.dtype))
 
     out = (
         df_X,
@@ -367,6 +384,28 @@ def _validate_api_params(  # noqa: C901
     return out
 
 
+def _check_configure_joblib_backend() -> None:
+    """Check and configure if we are running in a joblib parallel backend."""
+    backend, _ = get_active_backend()
+
+    # If we are in an HTCondor backend, we need to set the context function to
+    # set the global configuration variables from both julearn and scikit-
+    # learn, as these are not automatically propagated to the workers in
+    # non-shared memory backends.
+    if backend.__class__.__name__ == "_HTCondorBackend":
+        if hasattr(backend, "_context_func"):
+            backend._context_func = _joblib_htcondor_context_func(
+                current_func=backend._context_func
+            )
+        else:
+            raise_error(
+                "Running in an HTCondor backend, but it does not have a "
+                "_context_func attribute. Global configuration variables "
+                "will not be propagated to the workers. Please update your "
+                "joblib-htcondor package to version 0.2.3 or higher."
+            )
+
+
 def run_cross_validation(
     X: list[str],  # noqa: N803
     y: str,
@@ -563,7 +602,7 @@ def run_cross_validation(
         classifier=problem_type == "classification",
         include_final_model=include_final_model,
     )
-    logger.info(f"Using outer CV scheme {cv_outer}")
+    logger.info(__("Using outer CV scheme {cv_outer}", cv_outer=cv_outer))
 
     groups_needed = check_consistency(df_y, cv, groups, problem_type)  # type: ignore
 
@@ -598,6 +637,8 @@ def run_cross_validation(
         else:
             _sklearn_deprec_fit_params["fit_params"] = fit_params
 
+    _check_configure_joblib_backend()
+
     scores = cross_validate(
         pipeline,
         df_X,
@@ -820,6 +861,8 @@ def run_fit(
     else:
         _sklearn_deprec_fit_params["fit_params"] = fit_params
 
+    _check_configure_joblib_backend()
+
     logger.info("Fitting final model")
     pipeline.fit(df_X, df_y, **fit_params)
 

julearn-0.3.6.dev15/julearn/config.py

@@ -0,0 +1,118 @@
+"""Global config module."""
+
+# Authors: Federico Raimondo <f.raimondo@fz-juelich.de>
+# License: AGPL
+
+from typing import Any
+
+from .utils import logger, raise_error
+from .utils.logging import DelayedFmtMessage as __
+
+
+_global_config = {}
+_global_config["MAX_X_WARNS"] = 5000
+_global_config["max_x_logs"] = 100
+_global_config["disable_x_check"] = False
+_global_config["disable_xtypes_check"] = False
+_global_config["disable_x_verbose"] = False
+_global_config["disable_xtypes_verbose"] = False
+_global_config["enable_parallel_column_transformers"] = False
+_global_config["enable_auto_escape_parenthesis"] = True
+
+
+def set_config(key: str, value: Any) -> None:
+    """Set a global config value.
+
+    Parameters
+    ----------
+    key : str
+        The key to set.
+    value : Any
+        The value to set.
+
+    """
+    if key not in _global_config:
+        raise_error(f"Global config {key} does not exist")
+    logger.info(
+        __("Setting global config {key} to {value}", key=key, value=value)
+    )
+    _global_config[key] = value
+
+
+def get_config(key: str) -> Any:
+    """Get a global config value.
+
+    Parameters
+    ----------
+    key : str
+        The key to get.
+
+    Returns
+    -------
+    Any
+        The value of the key.
+
+    """
+    return _global_config.get(key, None)
+
+
+def _joblib_htcondor_context_func(current_func=None) -> None:
+    """Create a function to seet the config variables.
+
+    Parameters
+    ----------
+    current_func : function
+        The current function that was set in the joblib context. If not None,
+        it will be called before setting the config variables (default None).
+
+    Returns
+    -------
+    function
+        A function to set the config variables.
+
+    """
+    from copy import deepcopy
+    from functools import partial
+
+    import sklearn
+
+    # Copy the global config
+    _config = {}
+    _config["julearn_config"] = deepcopy(_global_config)
+
+    _config["sklearn_config"] = sklearn.get_config()
+
+    # add the logging configuration
+    level = logger.level
+    fmt = None
+    if len(logger.handlers) > 0:
+        fmt = logger.handlers[0].formatter._fmt  # type: ignore
+
+    _config["julearn_logging_config"] = {
+        "level": level,
+        "output_format": fmt,
+    }
+
+    # Log the current function, if any, to be called before setting the config
+    _config["jht_current_func"] = current_func
+
+    def _set_context_vars(**kwargs):
+        # First, call the previous function
+        jht_current_func = kwargs.pop("jht_current_func")
+        if jht_current_func is not None:
+            jht_current_func()
+
+        # Julearn then sets the config variables. If user has set any variable
+        # in the context function, current julearn state should override this
+        # setting.
+        for key, value in kwargs.items():
+            if key == "sklearn_config":
+                sklearn.set_config(**value)
+            elif key == "julearn_logging_config":
+                from julearn.utils import configure_logging
+                configure_logging(**value)
+            elif key == "julearn_config":
+                for k, v in value.items():
+                    set_config(k, v)
+
+    return partial(_set_context_vars, **_config)  # type: ignore

{julearn-0.3.5.dev140 → julearn-0.3.6.dev15}/julearn/model_selection/_optuna_searcher.py

@@ -6,6 +6,7 @@
 from typing import Any
 
 from ..utils import logger
+from ..utils.logging import DelayedFmtMessage as __
 from .available_searchers import _recreate_reset_copy, register_searcher
 
 
@@ -90,32 +91,57 @@ def _prepare_optuna_hyperparameters_distributions(
             if v[2] == "uniform":
                 if isinstance(v[0], int) and isinstance(v[1], int):
                     logger.info(
-                        f"Hyperparameter {k} is uniform integer "
-                        f"[{v[0]}, {v[1]}]"
+                        __(
+                            "Hyperparameter {k} is uniform integer "
+                            "[{v1}, {v2}]",
+                            k=k,
+                            v1=v[0],
+                            v2=v[1],
+                        )
                     )
                     out[k] = optd.IntDistribution(v[0], v[1], log=False)
                 else:
                     logger.info(
-                        f"Hyperparameter {k} is uniform float [{v[0]}, {v[1]}]"
+                        __(
+                            "Hyperparameter {k} is uniform float [{v1}, {v2}]",
+                            k=k,
+                            v1=v[0],
+                            v2=v[1],
+                        )
                     )
                     out[k] = optd.FloatDistribution(v[0], v[1], log=False)
             elif v[2] == "log-uniform":
                 if isinstance(v[0], int) and isinstance(v[1], int):
                     logger.info(
-                        f"Hyperparameter {k} is log-uniform int "
-                        f"[{v[0]}, {v[1]}]"
+                        __(
+                            "Hyperparameter {k} is log-uniform int "
+                            "[{v1}, {v2}]",
+                            k=k,
+                            v1=v[0],
+                            v2=v[1],
+                        )
                     )
                     out[k] = optd.IntDistribution(v[0], v[1], log=True)
                 else:
                     logger.info(
-                        f"Hyperparameter {k} is log-uniform float "
-                        f"[{v[0]}, {v[1]}]"
+                        __(
+                            "Hyperparameter {k} is log-uniform float "
+                            "[{v1}, {v2}]",
+                            k=k,
+                            v1=v[0],
+                            v2=v[1],
+                        )
                     )
                     out[k] = optd.FloatDistribution(v[0], v[1], log=True)
             elif v[2] == "categorical":
                 logger.info(
-                    f"Hyperparameter {k} is categorical with 2 "
-                    f"options: [{v[0]} and {v[1]}]"
+                    __(
+                        "Hyperparameter {k} is categorical with 2 "
+                        "options: [{v1} and {v2}]",
+                        k=k,
+                        v1=v[0],
+                        v2=v[1],
+                    )
                 )
                 out[k] = optd.CategoricalDistribution((v[0], v[1]))
             else:
@@ -125,9 +151,11 @@ def _prepare_optuna_hyperparameters_distributions(
             and isinstance(v[-1], str)
             and v[-1] == "categorical"
         ):
-            logger.info(f"Hyperparameter {k} is categorical [{v[:-1]}]")
+            logger.info(
+                __("Hyperparameter {k} is categorical [{v}]", k=k, v=v[:-1])
+            )
             out[k] = optd.CategoricalDistribution(v[:-1])
         else:
-            logger.info(f"Hyperparameter {k} as is {v}")
+            logger.info(__("Hyperparameter {k} as is {v}", k=k, v=v))
             out[k] = v
     return out

{julearn-0.3.5.dev140 → julearn-0.3.6.dev15}/julearn/model_selection/_skopt_searcher.py

@@ -6,6 +6,7 @@
 from typing import Any
 
 from ..utils import logger
+from ..utils.logging import DelayedFmtMessage as __
 from .available_searchers import _recreate_reset_copy, register_searcher
 
 
@@ -65,22 +66,39 @@ def _prepare_skopt_hyperparameters_distributions(
             prior = v[2]
             if prior == "categorical":
                 logger.info(
-                    f"Hyperparameter {k} is categorical with 2 "
-                    f"options: [{v[0]} and {v[1]}]"
+                    __(
+                        "Hyperparameter {k} is categorical with 2 "
+                        "options: [{v1} and {v2}]",
+                        k=k,
+                        v1=v[0],
+                        v2=v[1],
+                    )
                 )
                 out[k] = sksp.Categorical(v[:-1])
             elif isinstance(v[0], int) and isinstance(v[1], int):
                 logger.info(
-                    f"Hyperparameter {k} is {prior} integer [{v[0]}, {v[1]}]"
+                    __(
+                        "Hyperparameter {k} is {prior} integer [{v1}, {v2}]",
+                        k=k,
+                        prior=prior,
+                        v1=v[0],
+                        v2=v[1],
+                    )
                 )
                 out[k] = sksp.Integer(v[0], v[1], prior=prior)
             elif isinstance(v[0], float) and isinstance(v[1], float):
                 logger.info(
-                    f"Hyperparameter {k} is {prior} float [{v[0]}, {v[1]}]"
+                    __(
+                        "Hyperparameter {k} is {prior} float [{v1}, {v2}]",
+                        k=k,
+                        prior=prior,
+                        v1=v[0],
+                        v2=v[1],
+                    )
                 )
                 out[k] = sksp.Real(v[0], v[1], prior=prior)
             else:
-                logger.info(f"Hyperparameter {k} as is {v}")
+                logger.info(__("Hyperparameter {k} as is {v}", k=k, v=v))
                 out[k] = v
         elif (
             isinstance(v, tuple)
@@ -89,6 +107,6 @@ def _prepare_skopt_hyperparameters_distributions(
         ):
             out[k] = sksp.Categorical(v[:-1])
         else:
-            logger.info(f"Hyperparameter {k} as is {v}")
+            logger.info(__("Hyperparameter {k} as is {v}", k=k, v=v))
             out[k] = v
     return out

{julearn-0.3.5.dev140 → julearn-0.3.6.dev15}/julearn/model_selection/available_searchers.py

@@ -9,7 +9,14 @@ from typing import Optional, Union
 
 from sklearn.model_selection import GridSearchCV, RandomizedSearchCV
 
-from julearn.utils.logging import logger, raise_error, warn_with_log
+from julearn.utils.logging import (
+    DelayedFmtMessage as __,
+)
+from julearn.utils.logging import (
+    logger,
+    raise_error,
+    warn_with_log,
+)
 
 
 _available_searchers = {
@@ -125,7 +132,12 @@ def register_searcher(
                 "Set `overwrite=True` in case you want to "
                 "overwrite existing searchers."
             )
-    logger.info(f"Registering new searcher: {searcher_name}")
+    logger.info(
+        __(
+            "Registering new searcher: {searcher_name}",
+            searcher_name=searcher_name,
+        )
+    )
     _available_searchers[searcher_name] = {
         "class": searcher,
         "params_attr": params_attr,

{julearn-0.3.5.dev140 → julearn-0.3.6.dev15}/julearn/models/available_models.py

@@ -47,6 +47,7 @@ from sklearn.naive_bayes import (
 from sklearn.svm import SVC, SVR
 
 from ..utils import logger, raise_error, warn_with_log
+from ..utils.logging import DelayedFmtMessage as __
 from ..utils.typing import ModelLike
 from .dynamic import DynamicSelection
 
@@ -256,15 +257,23 @@ def register_model(
                         )
 
                     logger.info(
-                        f"registering model named {model_name} "
-                        f"with problem_type {problem_type}"
+                        __(
+                            "registering model named {model_name} "
+                            "with problem_type {problem_type}",
+                            model_name=model_name,
+                            problem_type=problem_type,
+                        )
                     )
 
                 _available_models[model_name][problem_type] = cls
             else:
                 logger.info(
-                    f"registering model named {model_name} "
-                    f"with problem_type {problem_type}"
+                    __(
+                        "registering model named {model_name} "
+                        "with problem_type {problem_type}",
+                        model_name=model_name,
+                        problem_type=problem_type,
+                    )
                 )
                 _available_models[model_name] = {problem_type: cls}