py-automl-lib 2.2__tar.gz → 2.2.2__tar.gz

{py-automl-lib-2.2 → py_automl_lib-2.2.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: py-automl-lib
-Version: 2.2
+Version: 2.2.2
 Summary: Python package for automated hyperparameter-optimization of common machine-learning algorithms
 Author: Owen O'D
 Project-URL: Homepage, https://github.com/owenodriscoll/AutoML
@@ -21,6 +21,8 @@ Requires-Dist: SQLAlchemy
 Provides-Extra: shap
 Requires-Dist: shap; extra == "shap"
 
+[![PyPI version](https://img.shields.io/pypi/v/py-automl-lib.svg?color=4c1)](https://pypi.org/project/py-automl-lib/)
+
 # automl: Automated Machine Learning
 ## Intro
 automl is a python project focussed on automating much of the machine learning efforts encountered in zero-dimensional regression and classification (and thus not multidimensional data such as for a CNN). It relies on existing Python packages Sci-Kit Learn, Optuna and model specific packages LightGBM, CatBoost and XGBoost.

{py-automl-lib-2.2 → py_automl_lib-2.2.2}/README.md

@@ -1,3 +1,5 @@
+[![PyPI version](https://img.shields.io/pypi/v/py-automl-lib.svg?color=4c1)](https://pypi.org/project/py-automl-lib/)
+
 # automl: Automated Machine Learning
 ## Intro
 automl is a python project focussed on automating much of the machine learning efforts encountered in zero-dimensional regression and classification (and thus not multidimensional data such as for a CNN). It relies on existing Python packages Sci-Kit Learn, Optuna and model specific packages LightGBM, CatBoost and XGBoost.

{py-automl-lib-2.2 → py_automl_lib-2.2.2}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "py-automl-lib"
-version = "2.2"
+version = "2.2.2"
 authors = [
   { name="Owen O'D"},
 ]

{py-automl-lib-2.2 → py_automl_lib-2.2.2}/src/automl/automl.py

@@ -22,7 +22,7 @@ from sklearn.model_selection import train_test_split
 
 from .scalers_transformers import PcaChooser, PolyChooser, SplineChooser, ScalerChooser, \
     TransformerChooser, CategoricalChooser#, FourrierExpansion
-from misc.function_helper import FuncHelper
+from .utils.function_helper import FuncHelper
 
 # --------------- TODO LIST ---------------
 # FIXME add encoding for clustering of feature importance 
@@ -559,7 +559,7 @@ class AutomatedML:
 
             return 
 
-    def model_select_best(self, random_state_model_selection=None) -> AutomatedML:
+    def model_select_best(self, random_state_model_selection: int = None, performance_sign_positive: bool = True) -> AutomatedML:
         """
         This method is used to create estimator pipelines for all the models specified in models_to_assess
         attribute and store them in the estimators attribute of the class instance.
@@ -570,6 +570,13 @@ class AutomatedML:
         pipeline using the Pipeline class from scikit-learn library. Each pipeline per model is added to a list of
         pipelines, which is then assigned to the estimators attribute of the class instance.
 
+        Parameters
+        ----------
+        random_state_model_selection: int, None
+            integer used to set the random state for random weak model selection
+        performance_sign_positive: bool, True
+            boolean used to indicate whether performance values should be positive or negative only 
+
         Returns
         -------
         class instance.
@@ -580,9 +587,10 @@ class AutomatedML:
         for model_name in list(self._models_assess.keys()):
 
             # -- set randomness parameters for randomly selecting models (if self.n_weak_models > 0)
-            if type(random_state_model_selection) == type(None):
-                random_state_model_selection = self.random_state
-            random.seed(random_state_model_selection)
+            if random_state_model_selection == None:
+                random.seed(self.random_state)
+            else:
+                random.seed(random_state_model_selection)
 
             # -- reload relevant study. Sampler not reloaded here as no additional studies are performed
             study = optuna.create_study(
@@ -599,7 +607,17 @@ class AutomatedML:
 
             # -- select all trials associated with model
             df_trials = study.trials_dataframe()
-            df_trials_non_pruned = df_trials[df_trials.state == 'COMPLETE']
+
+            # -- remove trials returning NaN's or undesired_sign
+            if performance_sign_positive:
+                error_sign_condition = (df_trials.value >= 0)
+            else: 
+                error_sign_condition = (df_trials.value <= 0)
+
+            df_trials_non_pruned = df_trials[((df_trials.state == 'COMPLETE') | 
+                                                (df_trials.state == 'PRUNED')) & 
+                                                (np.isfinite(df_trials.value)) & 
+                                                (error_sign_condition) ]
 
             # -- ensure that selected number of weak models does not exceed `total completed trials` - `best trial`
             n_weak_models = self.n_weak_models
@@ -623,6 +641,10 @@ class AutomatedML:
             idx_remaining.remove(idx_best)
             idx_models = [idx_best] + random.sample(idx_remaining, n_weak_models)
 
+            # -- reset random state to pre-sampling state 
+            if random_state_model_selection == None:
+                random.seed(self.random_state)
+
             # -- name best and weaker models
             selected_models = [model_name+'_best']  + [model_name+'_'+str(i) for i in idx_models[1:]]
 
@@ -783,6 +805,18 @@ class AutomatedML:
     
 
     def apply(self):
+            """
+            Summary method that runs in sequence the following methods:
+
+            ```python
+            self.model_hyperoptimise()
+            self.model_select_best()
+            self.model_evaluate()
+            ```
+
+            These can also be called manually, for instance to skip an already performed hyperopimisation
+            """
+
             self.model_hyperoptimise()
             self.model_select_best()
             self.model_evaluate()
@@ -792,9 +826,9 @@ class AutomatedML:
     
     def model_feature_importance(self, n_train_points = 200, n_test_points = 200, cluster = True):
         """
-        NOTE DOES NOT WORK WITH NON-NUMERIC DATA
-        NOTE requires installation of the shap package
-            python3 -m pip install pyautoml[shap]
+        NOTE DOES NOT WORK WITH NON-NUMERIC DATA \n
+        NOTE requires installation of the shap package \n
+            `python3 -m pip install py-automl-lib[shap]`
 
         Evaluates feature importance using shapely values. The SHAP kernel explainer is trained on 
         the training data (or on the cluster thereof). Then the explainer calculates for the test 

{py-automl-lib-2.2 → py_automl_lib-2.2.2}/src/automl/scalers_transformers.py

@@ -20,7 +20,7 @@ import numpy as np
 import pandas as pd
 
 from typing import Union
-from misc.function_helper import FuncHelper
+from .utils.function_helper import FuncHelper
 
 
 def decorator_report(variable, to_return_self: bool = False):

{py-automl-lib-2.2 → py_automl_lib-2.2.2}/src/py_automl_lib.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: py-automl-lib
-Version: 2.2
+Version: 2.2.2
 Summary: Python package for automated hyperparameter-optimization of common machine-learning algorithms
 Author: Owen O'D
 Project-URL: Homepage, https://github.com/owenodriscoll/AutoML
@@ -21,6 +21,8 @@ Requires-Dist: SQLAlchemy
 Provides-Extra: shap
 Requires-Dist: shap; extra == "shap"
 
+[![PyPI version](https://img.shields.io/pypi/v/py-automl-lib.svg?color=4c1)](https://pypi.org/project/py-automl-lib/)
+
 # automl: Automated Machine Learning
 ## Intro
 automl is a python project focussed on automating much of the machine learning efforts encountered in zero-dimensional regression and classification (and thus not multidimensional data such as for a CNN). It relies on existing Python packages Sci-Kit Learn, Optuna and model specific packages LightGBM, CatBoost and XGBoost.

{py-automl-lib-2.2 → py_automl_lib-2.2.2}/src/py_automl_lib.egg-info/SOURCES.txt

@@ -1,6 +1,7 @@
 LICENSE
 README.md
 pyproject.toml
+src/__init__.py
 src/automl/__init__.py
 src/automl/auto_classification.py
 src/automl/auto_regression.py
@@ -8,9 +9,9 @@ src/automl/automl.py
 src/automl/classifiers.py
 src/automl/regressors.py
 src/automl/scalers_transformers.py
-src/misc/__init__.py
-src/misc/function_helper.py
-src/misc/misc.py
+src/automl/utils/__init__.py
+src/automl/utils/function_helper.py
+src/automl/utils/misc.py
 src/py_automl_lib.egg-info/PKG-INFO
 src/py_automl_lib.egg-info/SOURCES.txt
 src/py_automl_lib.egg-info/dependency_links.txt

py_automl_lib-2.2.2/src/py_automl_lib.egg-info/top_level.txt

@@ -0,0 +1,2 @@
+__init__
+automl

py-automl-lib-2.2/src/py_automl_lib.egg-info/top_level.txt

@@ -1,2 +0,0 @@
-automl
-misc