updating organization of deephyper.analysis

deephyper/__init__.py

@@ -1,5 +1,5 @@
 """
-DeepHyper's software architecture is designed to be modular and extensible. It is built on top of the following main sub-packages:
+DeepHyper's software architecture is designed to be modular and extensible. It is built on top of the following main subpackages:
 
 * :mod:`deephyper.ensemble`: Tools to build ensembles of neural networks with uncertainty quantification.
 * :mod:`deephyper.nas`: Tools to define neural architecture search space and evaluation strategy.
@@ -12,6 +12,7 @@
 DeepHyper installation requires **Python >= 3.7**.
 
 """
+
 import warnings
 from deephyper.__version__ import __version__, __version_suffix__  # noqa: F401
 

deephyper/analysis/__init__.py

@@ -1,7 +1,8 @@
 """
-This analysis sub-package contains modules to analyze the results deephyper.
+This analysis subpackage contains modules to analyze results returned by deephyper.
 """
 
 from ._rank import rank
+from ._matplotlib import figure_size, update_matplotlib_rc
 
-__all__ = ["rank"]
+__all__ = ["rank", "figure_size", "update_matplotlib_rc"]

deephyper/analysis/hps/__init__.py

@@ -0,0 +1,18 @@
+"""Subpackage for hyperparameter search analysis.
+"""
+
+from ._hps import (
+    filter_failed_objectives,
+    parameters_at_max,
+    plot_search_trajectory_single_objective_hps,
+    plot_worker_utilization,
+    read_results_from_csv,
+)
+
+__all__ = [
+    "filter_failed_objectives",
+    "parameters_at_max",
+    "plot_search_trajectory_single_objective_hps",
+    "plot_worker_utilization",
+    "read_results_from_csv",
+]

deephyper/analysis/hpo.py → deephyper/analysis/hps/_hps.py

@@ -10,10 +10,24 @@
 from matplotlib.ticker import MaxNLocator
 
 from deephyper.analysis import rank
-from deephyper.analysis._paxplot import pax_parallel
+from deephyper.analysis.hps._paxplot import pax_parallel
 
 
-def filter_failed_objectives(df: pd.DataFrame) -> Tuple[pd.DataFrame, pd.DataFrame]:
+def read_results_from_csv(file_path: str) -> pd.DataFrame:
+    """Read the results of a Hyperparameter Search from a CSV file.
+
+    Args:
+        file_path (str): the path to the CSV file.
+
+    Returns:
+        pd.DataFrame: the results of a Hyperparameter Search.
+    """
+    return pd.read_csv(file_path, index_col=None)
+
+
+def filter_failed_objectives(
+    df: pd.DataFrame,
+) -> Tuple[pd.DataFrame, pd.DataFrame]:
     """Filter out lines from the DataFrame with failed objectives.
 
     Args:
@@ -24,15 +38,15 @@ def filter_failed_objectives(df: pd.DataFrame) -> Tuple[pd.DataFrame, pd.DataFra
     """
     # Single-Objective
     if "objective" in df.columns:
-        if pd.api.types.is_string_dtype(df.objective):
-            mask = df.objective.str.startswith("F")
+        if pd.api.types.is_string_dtype(df["objective"]):
+            mask = df["objective"].str.startswith("F")
 
             df_with_failures = df[mask]
 
             df_without_failures = df[~mask]
-            df_without_failures.loc[
-                :, "objective"
-            ] = df_without_failures.objective.astype(float)
+            df_without_failures.loc[:, "objective"] = df_without_failures[
+                "objective"
+            ].astype(float)
         else:
             df_without_failures = df
             df_with_failures = df[np.zeros(len(df), dtype=bool)]
@@ -77,31 +91,32 @@ def parameters_at_max(
     return config, value
 
 
-def plot_search_trajectory_single_objective_hpo(
-    results, show_failures: bool = True, ax=None, **kwargs
+def plot_search_trajectory_single_objective_hps(
+    results, show_failures: bool = True, column="objective", ax=None, **kwargs
 ):
     """Plot the search trajectory of a Single-Objective Hyperparameter Search.
 
     Args:
         results (pd.DataFrame): the results of a Hyperparameter Search.
         show_failures (bool, optional): whether to show the failed objectives. Defaults to ``True``.
+        column (str, optional): the column to use for the y-axis of the plot. Defaults to ``"objective"``.
         ax (matplotlib.pyplot.axes): the axes to use for the plot.
 
     Returns:
         (matplotlib.pyplot.figure, matplotlib.pyplot.axes): the figure and axes of the plot.
     """
 
-    if results.objective.dtype != np.float64:
+    if results[column].dtype != np.float64:
         x = np.arange(len(results))
-        mask_failed = np.where(results.objective.str.startswith("F"))[0]
-        mask_success = np.where(~results.objective.str.startswith("F"))[0]
+        mask_failed = np.where(results[column].str.startswith("F"))[0]
+        mask_success = np.where(~results[column].str.startswith("F"))[0]
         x_success, x_failed = x[mask_success], x[mask_failed]
-        y_success = results.objective[mask_success].astype(float)
+        y_success = results[column][mask_success].astype(float)
     else:
         x = np.arange(len(results))
         x_success = x
         x_failed = np.array([])
-        y_success = results.objective
+        y_success = results[column]
 
     y_min, y_max = y_success.min(), y_success.max()
     y_min = y_min - 0.05 * (y_max - y_min)
@@ -223,7 +238,6 @@ def plot_worker_utilization(
         ax.set_ylabel("Utilization")
     else:
         ax.set_ylabel("Active Workers")
-    ax.legend()
     ax.grid(True)
     ax.set_xlim(x.min(), x.max())
 

deephyper/core/cli/_new_problem.py

@@ -2,14 +2,15 @@
 Create a DeepHyper Problem
 --------------------------
 
-Command line to create a new problem sub-package in a DeepHyper projet package.
+Command line to create a new problem subpackage in a DeepHyper projet package.
 
 It can be used with:
 
 .. code-block:: console
 
     $ deephyper new-problem hps problem_name
 """
+
 import glob
 import os
 import pathlib

deephyper/evaluator/__init__.py

@@ -1,5 +1,5 @@
 """
-This evaluator sub-package provides a common interface to execute isolated tasks with different parallel backends and system properties. This interface is used by search algorithm to perform black-box optimization (the black-box being represented by the ``run``-function).
+This evaluator subpackage provides a common interface to execute isolated tasks with different parallel backends and system properties. This interface is used by search algorithm to perform black-box optimization (the black-box being represented by the ``run``-function).
 An ``Evaluator``, when instanciated, is bound to a ``run``-function which takes as first argument a dictionnary and optionally has other keyword-arguments. The ``run``-function has to return a Python serializable value (under ``pickle`` protocol). In it's most basic form the return value is a ``float``.
 
 An example ``run``-function is:

deephyper/evaluator/storage/__init__.py

@@ -1,5 +1,6 @@
-"""This sub-package provides an interface to implement new storage clients. The base class defining this interface is the :class:`deephyper.evaluator.storage.Storage`. A storage in our langage is a synonym of memory. Different databases  or memory systems can be used through this interface (e.g., key-value storage, relational database, etc.).
+"""This subpackage provides an interface to implement new storage clients. The base class defining this interface is the :class:`deephyper.evaluator.storage.Storage`. A storage in our langage is a synonym of memory. Different databases  or memory systems can be used through this interface (e.g., key-value storage, relational database, etc.).
 """
+
 from deephyper.evaluator.storage._storage import Storage
 from deephyper.evaluator.storage._memory_storage import MemoryStorage
 

deephyper/nas/__init__.py

@@ -1,9 +1,10 @@
-"""This sub-package his dedicated to the definition of neural architecture search space and evaluation strategy. The implementation is using Tensorflow 2.X and Keras API. The main concepts are:
+"""This subpackage his dedicated to the definition of neural architecture search space and evaluation strategy. The implementation is using Tensorflow 2.X and Keras API. The main concepts are:
 * :class:`deephyper.nas.KSearchSpace`: An object to define a search space of neural architectures.
-* :mod:`deephyper.nas.run`: A sub-package to define the evaluation strategy of a neural architecture (e.g., training procedure).
-* :mod:`deephyper.nas.operation`: A sub-package to define operations of the neural architecture search space.
-* :mod:`deephyper.nas.node`: A sub-package to define nodes of the neural architecture search space which is represented as a direct acyclic graph.
+* :mod:`deephyper.nas.run`: A subpackage to define the evaluation strategy of a neural architecture (e.g., training procedure).
+* :mod:`deephyper.nas.operation`: A subpackage to define operations of the neural architecture search space.
+* :mod:`deephyper.nas.node`: A subpackage to define nodes of the neural architecture search space which is represented as a direct acyclic graph.
 """
+
 from ._nx_search_space import NxSearchSpace
 from ._keras_search_space import KSearchSpace
 

deephyper/nas/run/__init__.py

@@ -1,5 +1,6 @@
-"""The :mod:`deephyper.nas.run` sub-package provides a set of functions which can evaluates configurations generated by search algorithms of DeepHyper.
+"""The :mod:`deephyper.nas.run` subpackage provides a set of functions which can evaluates configurations generated by search algorithms of DeepHyper.
 """
+
 from ._run_base_trainer import run_base_trainer
 from ._run_distributed_base_trainer import run_distributed_base_trainer
 from ._run_debug_arch import run_debug_arch

deephyper/problem/__init__.py

@@ -1,8 +1,9 @@
-"""This sub-package provides tools to define hyperparameter and neural architecture search problems. Some features of this module are based on the `ConfigSpace <https://automl.github.io/ConfigSpace/master/>`_ project. The main classes provided by this module are:
+"""This subpackage provides tools to define hyperparameter and neural architecture search problems. Some features of this module are based on the `ConfigSpace <https://automl.github.io/ConfigSpace/master/>`_ project. The main classes provided by this module are:
 
 - :class:`deephyper.problem.HpProblem`: A class to define a hyperparameter search problem.
 - :class:`deephyper.problem.NaProblem`: A class to define a neural architecture search problem.
 """
+
 from ConfigSpace import *  # noqa: F401, F403
 
 from ._hyperparameter import HpProblem

deephyper/search/__init__.py

@@ -1,4 +1,4 @@
-"""This sub-package provides an interface to implement new search algorithms as well as some already implemented search algorithms. One module of this sub-package is specialized for hyperparameter optimization algorithms ``deephyper.search.hps`` and an other is specialized for neural architecture search ``deephyper.search.nas``.
+"""This subpackage provides an interface to implement new search algorithms as well as some already implemented search algorithms. One module of this subpackage is specialized for hyperparameter optimization algorithms ``deephyper.search.hps`` and an other is specialized for neural architecture search ``deephyper.search.nas``.
 
 The :class:`deephyper.search.Search` class provides the generic interface of a search.
 

deephyper/search/hps/__init__.py

@@ -1,4 +1,4 @@
-"""Sub-package for hyperparameter search algorithms.
+"""Subpackage for hyperparameter search algorithms.
 
 .. warning:: All search algorithms are MAXIMIZING the objective function. If you want to MINIMIZE the objective function, you have to return the negative of you objective.
 """

deephyper/search/nas/__init__.py

@@ -1,7 +1,8 @@
-"""Sub-package for neural architecture search algorithms.
+"""Subpackage for neural architecture search algorithms.
 
 .. warning:: All search algorithms are MAXIMIZING the objective function. If you want to MINIMIZE the objective function, you have to return the negative of you objective.
 """
+
 from deephyper.search.nas._base import NeuralArchitectureSearch
 from deephyper.search.nas._regevo import RegularizedEvolution
 from deephyper.search.nas._agebo import AgEBO

deephyper/sklearn/__init__.py

@@ -1,2 +1,2 @@
-"""Sub-package providing tools for automl.
+"""Subpackage providing tools for automl.
 """

deephyper/test/__init__.py

@@ -1,4 +1,4 @@
-"""Sub-package dedicated to reusable testing tools for DeepHyper"""
+"""Subpackage dedicated to reusable testing tools for DeepHyper"""
 
 from ._command import run
 from ._parse_result import parse_result