ENH: Migrate to model_selection.load_classifier

README.md

@@ -90,18 +90,18 @@ ORCA-python includes sample datasets with pre-partitioned train/test splits usin
     },
     "configurations": {
         "SVM": {
-            "classifier": "sklearn.svm.SVC",
+            "classifier": "SVC",
             "parameters": {
                 "C": [0.001, 0.1, 1, 10, 100],
                 "gamma": [0.1, 1, 10]
             }
         },
         "SVMOP": {
-            "classifier": "orca_python.classifiers.OrdinalDecomposition",
+            "classifier": "OrdinalDecomposition",
             "parameters": {
                 "dtype": "ordered_partitions",
                 "decision_method": "frank_hall",
-                "base_classifier": "sklearn.svm.SVC",
+                "base_classifier": "SVC",
                 "parameters": {
                     "C": [0.01, 0.1, 1, 10],
                     "gamma": [0.01, 0.1, 1, 10],
@@ -144,7 +144,7 @@ Controls global experiment parameters.
 
 Defines classifiers and their hyperparameters for GridSearchCV. Each configuration has a name and consists of:
 
-- **`classifier`**: scikit-learn path or built-in ORCA-python classifier
+- **`classifier`**: scikit-learn or built-in ORCA-python classifier
 - **`parameters`**: hyperparameters for grid search (nested for ensemble methods)
 
 ## Running Experiments

orca_python/classifiers/OrdinalDecomposition.py

@@ -5,7 +5,7 @@
 from sklearn.utils._param_validation import StrOptions
 from sklearn.utils.validation import check_array, check_is_fitted, check_X_y
 
-from orca_python.utilities import load_classifier
+from orca_python.model_selection import load_classifier
 
 
 class OrdinalDecomposition(BaseEstimator, ClassifierMixin):
@@ -110,7 +110,7 @@ def __init__(
         self,
         dtype="ordered_partitions",
         decision_method="frank_hall",
-        base_classifier="sklearn.linear_model.LogisticRegression",
+        base_classifier="LogisticRegression",
         parameters={},
     ):
         self.dtype = dtype
@@ -161,7 +161,9 @@ def fit(self, X, y):
         # Fitting n_targets - 1 classifiers
         for n in range(len(class_labels[0, :])):
 
-            estimator = load_classifier(self.base_classifier, self.parameters)
+            estimator = load_classifier(
+                self.base_classifier, param_grid=self.parameters
+            )
             estimator.fit(
                 X[np.where(class_labels[:, n] != 0)],
                 np.ravel(class_labels[np.where(class_labels[:, n] != 0), n].T),

orca_python/classifiers/tests/test_ordinal_decomposition.py

@@ -24,7 +24,7 @@ def test_ordinal_decomposition(X, y):
     classifier = OrdinalDecomposition(
         dtype="ordered_partitions",
         decision_method="frank_hall",
-        base_classifier="sklearn.svm.SVC",
+        base_classifier="SVC",
         parameters={"C": 1.0, "gamma": "scale", "probability": True},
     )
 

orca_python/configurations/full_functionality_test.json

@@ -18,18 +18,18 @@
     },
     "configurations": {
         "SVM": {
-            "classifier": "sklearn.svm.SVC",
+            "classifier": "SVC",
             "parameters": {
                 "C": [0.001, 0.1, 1, 10, 100],
                 "gamma": [0.1, 1, 10]
             }
         },
         "SVMOP": {
-            "classifier": "orca_python.classifiers.OrdinalDecomposition",
+            "classifier": "OrdinalDecomposition",
             "parameters": {
                 "dtype": "ordered_partitions",
                 "decision_method": "frank_hall",
-                "base_classifier": "sklearn.svm.SVC",
+                "base_classifier": "SVC",
                 "parameters": {
                     "C": [0.01, 0.1, 1, 10],
                     "gamma": [0.01, 0.1, 1, 10],
@@ -38,11 +38,11 @@
             }
         },
         "LR": {
-            "classifier": "orca_python.classifiers.OrdinalDecomposition",
+            "classifier": "OrdinalDecomposition",
             "parameters": {
                 "dtype": ["ordered_partitions", "one_vs_next"],
                 "decision_method": "exponential_loss",
-                "base_classifier": "sklearn.linear_model.LogisticRegression",
+                "base_classifier": "LogisticRegression",
                 "parameters": {
                     "solver": ["liblinear"],
                     "C": [0.01, 0.1, 1, 10],
@@ -52,19 +52,19 @@
             }
         },
         "REDSVM": {
-            "classifier": "orca_python.classifiers.REDSVM",
+            "classifier": "REDSVM",
             "parameters": {
                 "kernel": "rbf",
                 "degree": 3,
                 "gamma": 0.1,
                 "coef0": 0,
                 "C": 1,
                 "tol": 0.001,
-                "shrinking": 1
+                "shrinking": true
             }
         },
         "SVOREX": {
-            "classifier": "orca_python.classifiers.SVOREX",
+            "classifier": "SVOREX",
             "parameters": {
                 "kernel": "gaussian",
                 "C": [0.1, 1, 10],

orca_python/configurations/nnop_mae_test.json

@@ -11,7 +11,7 @@
     },
     "configurations": {
         "NNOP-MAE": {
-            "classifier": "orca_python.classifiers.NNOP",
+            "classifier": "NNOP",
             "parameters": {
                 "epsilon_init": 0.5,
                 "n_hidden": [5, 10, 20, 30, 40, 50],

orca_python/configurations/nnop_mze_test.json

@@ -11,7 +11,7 @@
 	},
 	"configurations": {
 		"NNOP-MZE": {
-			"classifier": "orca_python.classifiers.NNOP",
+			"classifier": "NNOP",
 			"parameters": {
                 "epsilon_init": 0.5,
                 "n_hidden": [5, 10, 20, 30, 40, 50],

orca_python/configurations/nnpom_mae_test.json

@@ -11,7 +11,7 @@
 	},
 	"configurations": {
 		"NNPOM-MAE": {
-			"classifier": "orca_python.classifiers.NNPOM",
+			"classifier": "NNPOM",
 			"parameters": {
                 "epsilon_init": 0.5,
                 "n_hidden": [5, 10, 20, 30, 40, 50],

orca_python/configurations/nnpom_mze_test.json

@@ -11,7 +11,7 @@
     },
     "configurations": {
         "NNPOM-MZE": {
-            "classifier": "orca_python.classifiers.NNPOM",
+            "classifier": "NNPOM",
             "parameters": {
                 "epsilon_init": 0.5,
                 "n_hidden": [5, 10, 20, 30, 40, 50],

orca_python/configurations/redsvm_test.json

@@ -11,7 +11,7 @@
     },
     "configurations": {
         "REDSVM": {
-            "classifier": "orca_python.classifiers.REDSVM",
+            "classifier": "REDSVM",
             "parameters": {
                 "kernel": "rbf",
                 "C": [0.001, 0.01, 0.1, 1, 10, 100, 1000],

orca_python/configurations/svmop_test.json

@@ -10,11 +10,11 @@
     },
     "configurations": {
         "SVMOP": {
-            "classifier": "orca_python.classifiers.OrdinalDecomposition",
+            "classifier": "OrdinalDecomposition",
             "parameters": {
                 "dtype": "ordered_partitions",
                 "decision_method": "frank_hall",
-                "base_classifier": "sklearn.svm.SVC",
+                "base_classifier": "SVC",
                 "parameters": {
                     "C": [0.1, 1, 10],
                     "gamma": [0.1, 1, 10],

orca_python/configurations/svorex_test.json

@@ -11,7 +11,7 @@
     },
     "configurations": {
         "SVOREX": {
-            "classifier": "orca_python.classifiers.SVOREX",
+            "classifier": "SVOREX",
             "parameters": {
                 "kernel": "gaussian",
                 "C": [0.001, 0.01, 0.1, 1, 10, 100, 1000],

orca_python/utilities/__init__.py

@@ -2,16 +2,12 @@
 
 from .utilities import (
     Utilities,
-    check_for_random_state,
     check_packages_version,
     get_key,
-    load_classifier,
 )
 
 __all__ = [
     "Utilities",
     "check_packages_version",
-    "load_classifier",
-    "check_for_random_state",
     "get_key",
 ]

orca_python/utilities/tests/test_utilities.py

@@ -9,7 +9,7 @@
 import pytest
 
 from orca_python.testing import TEST_DATASETS_DIR
-from orca_python.utilities import Utilities, load_classifier
+from orca_python.utilities import Utilities
 
 
 @pytest.fixture
@@ -130,127 +130,6 @@ def test_standardize_data(util):
     npt.assert_almost_equal(np.std(std_X_train), 1)
 
 
-def test_load_algorithm():
-    # Loading a method from within this framework
-    from orca_python.classifiers import OrdinalDecomposition
-
-    imported_class = load_classifier("orca_python.classifiers.OrdinalDecomposition")
-    npt.assert_equal(imported_class, OrdinalDecomposition)
-
-    # Loading a scikit-learn classifier
-    from sklearn.svm import SVC
-
-    imported_class = load_classifier("sklearn.svm.SVC")
-    npt.assert_equal(imported_class, SVC)
-
-    # Raising exceptions when the classifier cannot be loaded
-    with pytest.raises(ImportError):
-        load_classifier("sklearn.svm.SVC.submethod")
-    with pytest.raises(AttributeError):
-        load_classifier("sklearn.svm.SVCC")
-
-
-def test_check_params(util):
-    """Testing functionality of check_params method.
-
-    It will test the 3 different scenarios contemplated within the framework
-    for passing the configuration.
-
-    """
-    # Normal use of configuration file with a non nested method
-    util.configurations = {
-        "conf1": {
-            "classifier": "sklearn.svm.SVC",
-            "parameters": {
-                "C": [0.1, 1, 10],
-                "gamma": [0.1, 1, 100],
-                "probability": "True",
-            },
-        }
-    }
-
-    # Getting formatted_params and expected_params
-    util._check_params()
-    formatted_params = util.configurations["conf1"]["parameters"]
-
-    random_state = util.configurations["conf1"]["parameters"]["random_state"]
-    expected_params = {
-        "C": [0.1, 1, 10],
-        "gamma": [0.1, 1, 100],
-        "probability": ["True"],
-        "random_state": random_state,
-    }
-
-    npt.assert_equal(formatted_params, expected_params)
-
-    # Configuration file using an ensemble method
-    util.configurations = {
-        "conf2": {
-            "classifier": "orca_python.classifiers.OrdinalDecomposition",
-            "parameters": {
-                "dtype": "OrderedPartitions",
-                "base_classifier": "sklearn.svm.SVC",
-                "parameters": {"C": [1, 10], "gamma": [1, 10], "probability": ["True"]},
-            },
-        }
-    }
-
-    # Getting formatted_params and expected_params
-    util._check_params()
-    formatted_params = util.configurations["conf2"]["parameters"]
-
-    random_state = util.configurations["conf2"]["parameters"]["parameters"][0][
-        "random_state"
-    ]
-    expected_params = {
-        "dtype": ["OrderedPartitions"],
-        "base_classifier": ["sklearn.svm.SVC"],
-        "parameters": [
-            {"C": 1, "gamma": 1, "probability": True, "random_state": random_state},
-            {"C": 1, "gamma": 10, "probability": True, "random_state": random_state},
-            {"C": 10, "gamma": 1, "probability": True, "random_state": random_state},
-            {"C": 10, "gamma": 10, "probability": True, "random_state": random_state},
-        ],
-    }
-
-    # Ordering list of parameters from formatted_params to prevent inconsistencies
-    formatted_params["parameters"] = sorted(
-        formatted_params["parameters"], key=lambda k: k["C"]
-    )
-
-    npt.assert_equal(expected_params, formatted_params)
-
-    # Configuration file where it's not necessary to perform cross-validation
-    util.configurations = {
-        "conf3": {
-            "classifier": "orca_python.classifiers.OrdinalDecomposition",
-            "parameters": {
-                "dtype": "OrderedPartitions",
-                "base_classifier": "sklearn.svm.SVC",
-                "parameters": {"C": [1], "gamma": [1]},
-            },
-        }
-    }
-
-    # Getting formatted_params and expected_params
-    util._check_params()
-    formatted_params = util.configurations["conf3"]["parameters"]
-
-    random_state = util.configurations["conf3"]["parameters"]["parameters"][
-        "random_state"
-    ]
-    expected_params = {
-        "dtype": "OrderedPartitions",
-        "base_classifier": "sklearn.svm.SVC",
-        "parameters": {"C": 1, "gamma": 1, "random_state": random_state},
-    }
-
-    npt.assert_equal(formatted_params, expected_params)
-
-    # Resetting configurations to not interfere with other experiments
-    util.configurations = {}
-
-
 @pytest.fixture
 def main_folder():
     return Path(__file__).parent.parent.parent
@@ -279,15 +158,15 @@ def general_conf(dataset_folder):
 def configurations():
     return {
         "SVM": {
-            "classifier": "sklearn.svm.SVC",
+            "classifier": "SVC",
             "parameters": {"C": [0.001, 0.1, 1, 10, 100], "gamma": [0.1, 1, 10]},
         },
         "SVMOP": {
-            "classifier": "orca_python.classifiers.OrdinalDecomposition",
+            "classifier": "OrdinalDecomposition",
             "parameters": {
                 "dtype": "ordered_partitions",
                 "decision_method": "frank_hall",
-                "base_classifier": "sklearn.svm.SVC",
+                "base_classifier": "SVC",
                 "parameters": {
                     "C": [0.01, 0.1, 1, 10],
                     "gamma": [0.01, 0.1, 1, 10],