moreo
/
QuaPy
1
0
Fork 1
Code Issues Pull Requests Releases Wiki Activity

mergin and solving pytests

This commit is contained in:
Alejandro Moreo Fernandez 2025-10-06 12:03:31 +02:00
parent dbda25b09a
commit 3847db3838
8 changed files with 131 additions and 89 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
examples/15.composable_methods.py
View File

@ -22,6 +22,7 @@ data = qp.data.preprocessing.text2tfidf(
min_df = 5,
)
training, testing = data.train_test
Xtr, ytr = training.Xy
# We start by recovering PACC from its building blocks, a LeastSquaresLoss and
# a probabilistic ClassRepresentation. A 5-fold cross-validation is implemented
@ -46,7 +47,7 @@ pacc = ComposableQuantifier(
# Let's evaluate this quantifier.
print(f"Evaluating PACC: {pacc}")
pacc.fit(training)
pacc.fit(Xtr, ytr)
app = qp.protocol.APP(testing, sample_size=100, n_prevalences=21, repeats=1)
absolute_errors = qp.evaluation.evaluate(
model = pacc,
@ -70,7 +71,7 @@ model = ComposableQuantifier(
)
print(f"Evaluating {model}")
model.fit(training)
model.fit(Xtr, ytr)
absolute_errors = qp.evaluation.evaluate(
model = model,
protocol = app, # use the same protocol for evaluation
@ -125,7 +126,7 @@ grid_search = qp.model_selection.GridSearchQ(
error = "mae",
refit = False,
verbose = True,
).fit(training)
).fit(Xtr, ytr)
print(
f"Best hyper-parameters = {grid_search.best_params_}",
f"Best MAE = {grid_search.best_score_}",

3
quapy/method/__init__.py
View File

@ -27,7 +27,8 @@ AGGREGATIVE_METHODS = {
aggregative.KDEyML,
aggregative.KDEyCS,
aggregative.KDEyHD,
confidence.BayesianCC
# aggregative.OneVsAllAggregative,
confidence.BayesianCC,
}
BINARY_METHODS = {

10
quapy/method/aggregative.py
View File

@ -1406,18 +1406,20 @@ class OneVsAllAggregative(OneVsAllGeneric, AggregativeQuantifier):
`Gao and Sebastiani, 2016 <https://link.springer.com/content/pdf/10.1007/s13278-016-0327-z.pdf>`_.
:param binary_quantifier: a quantifier (binary) that will be employed to work on multiclass model in a
one-vs-all manner
one-vs-all manner (default PACC(LogitsticRegression()))
:param n_jobs: number of parallel workers
:param parallel_backend: the parallel backend for joblib (default "loky"); this is helpful for some quantifiers
(e.g., ELM-based ones) that cannot be run with multiprocessing, since the temp dir they create during fit will
is removed and no longer available at predict time.
"""
def __init__(self, binary_quantifier, n_jobs=None, parallel_backend='multiprocessing'):
def __init__(self, binary_quantifier=None, n_jobs=None, parallel_backend='multiprocessing'):
if binary_quantifier is None:
binary_quantifier = PACC()
assert isinstance(binary_quantifier, BaseQuantifier), \
f'{self.binary_quantifier} does not seem to be a Quantifier'
f'{binary_quantifier} does not seem to be a Quantifier'
assert isinstance(binary_quantifier, AggregativeQuantifier), \
f'{self.binary_quantifier} does not seem to be of type Aggregative'
f'{binary_quantifier} does not seem to be of type Aggregative'
self.binary_quantifier = binary_quantifier
self.n_jobs = qp._get_njobs(n_jobs)
self.parallel_backend = parallel_backend

117
quapy/method/composable.py
View File

@ -1,27 +1,28 @@
"""This module allows the composition of quantification methods from loss functions and feature transformations. This functionality is realized through an integration of the qunfold package: https://github.com/mirkobunse/qunfold."""
__install_istructions = """
from dataclasses import dataclass
from packaging.version import Version
from .base import BaseQuantifier
# what to display when an ImportError is thrown
_IMPORT_ERROR_MESSAGE = """qunfold, the back-end of quapy.method.composable, is not properly installed.
To fix this error, call:
pip install --upgrade pip setuptools wheel
pip install "jax[cpu]"
pip install "qunfold @ git+https://github.com/mirkobunse/qunfold@v0.1.5"
"""
__import_error_message = (
"qunfold, the back-end of quapy.method.composable, is not properly installed." + __install_istructions
)
__old_version_message = (
"The version of qunfold you have installed is not compatible with current quapy's version, "
"which requires qunfold>=0.1.5. " + __install_istructions
)
from packaging.version import Version
# try to import members of qunfold as members of this module
try:
import qunfold
from qunfold.quapy import QuaPyWrapper
from qunfold.base import BaseMixin
from qunfold.methods import AbstractMethod
from qunfold.sklearn import CVClassifier
from qunfold import (
LinearMethod, # methods
LeastSquaresLoss, # losses
BlobelLoss,
EnergyLoss,
@ -29,17 +30,20 @@ try:
CombinedLoss,
TikhonovRegularization,
TikhonovRegularized,
ClassTransformer, # transformers
HistogramTransformer,
DistanceTransformer,
KernelTransformer,
EnergyKernelTransformer,
LaplacianKernelTransformer,
GaussianKernelTransformer,
GaussianRFFKernelTransformer,
ClassRepresentation, # representations
HistogramRepresentation,
DistanceRepresentation,
KernelRepresentation,
EnergyKernelRepresentation,
LaplacianKernelRepresentation,
GaussianKernelRepresentation,
GaussianRFFKernelRepresentation,
)
except ImportError as e:
raise ImportError(_IMPORT_ERROR_MESSAGE) from e
__all__ = [ # control public members, e.g., for auto-documentation in sphinx; omit QuaPyWrapper
__all__ = [ # control public members, e.g., for auto-documentation in sphinx
"QUnfoldWrapper",
"ComposableQuantifier",
"CVClassifier",
"LeastSquaresLoss",
@ -49,17 +53,15 @@ try:
"CombinedLoss",
"TikhonovRegularization",
"TikhonovRegularized",
"ClassTransformer",
"HistogramTransformer",
"DistanceTransformer",
"KernelTransformer",
"EnergyKernelTransformer",
"LaplacianKernelTransformer",
"GaussianKernelTransformer",
"GaussianRFFKernelTransformer",
"ClassRepresentation",
"HistogramRepresentation",
"DistanceRepresentation",
"KernelRepresentation",
"EnergyKernelRepresentation",
"LaplacianKernelRepresentation",
"GaussianKernelRepresentation",
"GaussianRFFKernelRepresentation",
]
except ImportError as e:
raise ImportError(__import_error_message) from e
def check_compatible_qunfold_version():
@ -69,18 +71,54 @@ def check_compatible_qunfold_version():
# versions of qunfold <= 0.1.4 did not declare __version__ in the __init__.py but only in the setup.py
version_str = "0.1.4"
compatible = Version(version_str) >= Version("0.1.5")
installed_ver = Version(version_str)
required_ver = Version("0.1.5")
compatible = installed_ver.base_version == required_ver.base_version or installed_ver>=required_ver
return compatible
def ComposableQuantifier(loss, transformer, **kwargs):
@dataclass
class QUnfoldWrapper(BaseQuantifier,BaseMixin):
"""A thin wrapper for using qunfold methods in QuaPy.
Args:
_method: An instance of `qunfold.methods.AbstractMethod` to wrap.
Examples:
Here, we wrap an instance of ACC to perform a grid search with QuaPy.
>>> from qunfold import ACC
>>> qunfold_method = QUnfoldWrapper(ACC(RandomForestClassifier(obb_score=True)))
>>> quapy.model_selection.GridSearchQ(
>>> model = qunfold_method,
>>> param_grid = { # try both splitting criteria
>>> "representation__classifier__estimator__criterion": ["gini", "entropy"],
>>> },
>>> # ...
>>> )
"""
_method: AbstractMethod
def fit(self, X, y): # data is a qp.LabelledCollection
self._method.fit(X, y)
return self
def predict(self, X):
return self._method.predict(X)
def set_params(self, **params):
self._method.set_params(**params)
return self
def get_params(self, deep=True):
return self._method.get_params(deep)
def __str__(self):
return self._method.__str__()
def ComposableQuantifier(loss, representation, **kwargs):
"""A generic quantification / unfolding method that solves a linear system of equations.
This class represents any quantifier that can be described in terms of a loss function, a feature transformation, and a regularization term. In this implementation, the loss is minimized through unconstrained second-order minimization. Valid probability estimates are ensured through a soft-max trick by Bunse (2022).
Args:
loss: An instance of a loss class from `quapy.methods.composable`.
transformer: An instance of a transformer class from `quapy.methods.composable`.
representation: An instance of a representation class from `quapy.methods.composable`.
solver (optional): The `method` argument in `scipy.optimize.minimize`. Defaults to `"trust-ncg"`.
solver_options (optional): The `options` argument in `scipy.optimize.minimize`. Defaults to `{"gtol": 1e-8, "maxiter": 1000}`.
seed (optional): A random number generator seed from which a numpy RandomState is created. Defaults to `None`.
@ -92,12 +130,12 @@ def ComposableQuantifier(loss, transformer, **kwargs):
>>> ComposableQuantifier,
>>> TikhonovRegularized,
>>> LeastSquaresLoss,
>>> ClassTransformer,
>>> ClassRepresentation,
>>> )
>>> from sklearn.ensemble import RandomForestClassifier
>>> o_acc = ComposableQuantifier(
>>> TikhonovRegularized(LeastSquaresLoss(), 0.01),
>>> ClassTransformer(RandomForestClassifier(oob_score=True))
>>> ClassRepresentation(RandomForestClassifier(oob_score=True))
>>> )
Here, we perform hyper-parameter optimization with the ordinal ACC.
@ -105,7 +143,7 @@ def ComposableQuantifier(loss, transformer, **kwargs):
>>> quapy.model_selection.GridSearchQ(
>>> model = o_acc,
>>> param_grid = { # try both splitting criteria
>>> "transformer__classifier__estimator__criterion": ["gini", "entropy"],
>>> "representation__classifier__estimator__criterion": ["gini", "entropy"],
>>> },
>>> # ...
>>> )
@ -116,10 +154,7 @@ def ComposableQuantifier(loss, transformer, **kwargs):
>>> from sklearn.linear_model import LogisticRegression
>>> acc_lr = ComposableQuantifier(
>>> LeastSquaresLoss(),
>>> ClassTransformer(CVClassifier(LogisticRegression(), 10))
>>> ClassRepresentation(CVClassifier(LogisticRegression(), 10))
>>> )
"""
if not check_compatible_qunfold_version():
raise ImportError(__old_version_message)
return QuaPyWrapper(qunfold.GenericMethod(loss, transformer, **kwargs))
return QUnfoldWrapper(LinearMethod(loss, representation, **kwargs))

9
quapy/tests/test_datasets.py
View File

@ -15,8 +15,11 @@ class TestDatasets(unittest.TestCase):
return PCC(LogisticRegression(C=0.001, max_iter=100))
def _check_dataset(self, dataset):
train, test = dataset.reduce().train_test
q = self.new_quantifier()
print(f'testing method {q} in {dataset.name}...', end='')
if len(train)>500:
train = train.sampling(500)
q.fit(*dataset.training.Xy)
estim_prevalences = q.predict(dataset.test.instances)
self.assertTrue(F.check_prevalence_vector(estim_prevalences))
@ -42,7 +45,9 @@ class TestDatasets(unittest.TestCase):
self._check_dataset(dataset)
def test_twitter(self):
for dataset_name in TWITTER_SENTIMENT_DATASETS_TEST:
# all the datasets are contained in the same resource; if the first one
# works, there is no need to test for the rest
for dataset_name in TWITTER_SENTIMENT_DATASETS_TEST[:1]:
print(f'loading dataset {dataset_name}...', end='')
dataset = fetch_twitter(dataset_name, min_df=10)
dataset.stats()
@ -129,7 +134,7 @@ class TestDatasets(unittest.TestCase):
n_classes = train.n_classes
train = train.sampling(100, *F.uniform_prevalence(n_classes))
q = self.new_quantifier()
q.fit(train)
q.fit(*train.Xy)
self._check_samples(gen, q, max_samples_test=5)

3
quapy/tests/test_hierarchy.py
View File

@ -9,9 +9,8 @@ import inspect
class HierarchyTestCase(unittest.TestCase):
def test_aggregative(self):
lr = LogisticRegression()
for m in AGGREGATIVE_METHODS:
self.assertEqual(isinstance(m(lr), AggregativeQuantifier), True)
self.assertEqual(isinstance(m(), AggregativeQuantifier), True)
def test_inspect_aggregative(self):

12
quapy/tests/test_methods.py
View File

@ -4,6 +4,7 @@ import unittest
from sklearn.linear_model import LogisticRegression
import quapy as qp
from method.aggregative import OneVsAllAggregative
from quapy.method.aggregative import ACC
from quapy.method.meta import Ensemble
from quapy.method import AGGREGATIVE_METHODS, BINARY_METHODS, NON_AGGREGATIVE_METHODS
@ -16,21 +17,21 @@ from quapy.method.composable import (
ComposableQuantifier,
LeastSquaresLoss,
HellingerSurrogateLoss,
ClassTransformer,
HistogramTransformer,
ClassRepresentation,
HistogramRepresentation,
CVClassifier
)
COMPOSABLE_METHODS = [
ComposableQuantifier( # ACC
LeastSquaresLoss(),
ClassTransformer(CVClassifier(LogisticRegression()))
ClassRepresentation(CVClassifier(LogisticRegression()))
),
ComposableQuantifier( # HDy
HellingerSurrogateLoss(),
HistogramTransformer(
HistogramRepresentation(
3, # 3 bins per class
preprocessor = ClassTransformer(CVClassifier(LogisticRegression()))
preprocessor = ClassRepresentation(CVClassifier(LogisticRegression()))
)
),
]
@ -113,7 +114,6 @@ class TestMethods(unittest.TestCase):
self.assertTrue(check_prevalence_vector(estim_prevalences))
def test_composable(self):
from packaging.version import Version
if check_compatible_qunfold_version():
for dataset in TestMethods.datasets:
for q in COMPOSABLE_METHODS:

31
quapy/tests/test_modsel.py
View File

@ -39,31 +39,30 @@ class ModselTestCase(unittest.TestCase):
obtains the same optimal parameters
"""
q = PACC(LogisticRegression(random_state=1, max_iter=500))
q = PACC(LogisticRegression(random_state=1, max_iter=3000))
data = qp.datasets.fetch_reviews('imdb', tfidf=True, min_df=50).reduce(n_train=500, random_state=1)
data = qp.datasets.fetch_reviews('imdb', tfidf=True, min_df=50)
training, validation = data.training.split_stratified(0.7, random_state=1)
param_grid = {'classifier__C': np.logspace(-3,3,7)}
param_grid = {'classifier__C': np.logspace(-3,3,7), 'classifier__class_weight': ['balanced', None]}
app = APP(validation, sample_size=100, random_state=1)
def do_gridsearch(n_jobs):
print('starting model selection in sequential exploration')
tinit = time.time()
t_init = time.time()
modsel = GridSearchQ(
q, param_grid, protocol=app, error='mae', refit=False, timeout=-1, n_jobs=1, verbose=True
q, param_grid, protocol=app, error='mae', refit=False, timeout=-1, n_jobs=n_jobs, verbose=True
).fit(*training.Xy)
tend_seq = time.time()-tinit
best_c_seq = modsel.best_params_['classifier__C']
print(f'[done] took {tend_seq:.2f}s best C = {best_c_seq}')
t_end = time.time()-t_init
best_c = modsel.best_params_['classifier__C']
print(f'[done] took {t_end:.2f}s best C = {best_c}')
return t_end, best_c
print('starting model selection in parallel exploration')
tinit = time.time()
modsel = GridSearchQ(
q, param_grid, protocol=app, error='mae', refit=False, timeout=-1, n_jobs=-1, verbose=True
).fit(*training.Xy)
tend_par = time.time() - tinit
best_c_par = modsel.best_params_['classifier__C']
print(f'[done] took {tend_par:.2f}s best C = {best_c_par}')
tend_seq, best_c_seq = do_gridsearch(n_jobs=1)
tend_par, best_c_par = do_gridsearch(n_jobs=-1)
print(tend_seq, best_c_seq)
print(tend_par, best_c_par)
self.assertEqual(best_c_seq, best_c_par)
self.assertLess(tend_par, tend_seq)