improving plots debug

Fix PyPI: replace the direct extra dependency quapy[composable] with documentation on how to install through git

.github/workflows/ci.yml

@@ -20,15 +20,16 @@ jobs:
     env:
       QUAPY_TESTS_OMIT_LARGE_DATASETS: True
     steps:
-    - uses: actions/checkout@v3
+    - uses: actions/checkout@v4
     - name: Set up Python ${{ matrix.python-version }}
-      uses: actions/setup-python@v4
+      uses: actions/setup-python@v5
       with:
         python-version: ${{ matrix.python-version }}
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip setuptools wheel
-        python -m pip install -e .[bayes,composable,tests]
+        python -m pip install "qunfold @ git+https://github.com/mirkobunse/qunfold@v0.1.4"
+        python -m pip install -e .[bayes,tests]
     - name: Test with unittest
       run: python -m unittest
 
@@ -38,15 +39,18 @@ jobs:
     runs-on: ubuntu-latest
     if: github.ref == 'refs/heads/master'
     steps:
-    - uses: actions/checkout@v1
-    - name: Build documentation
-      uses: ammaraskar/sphinx-action@master
+    - uses: actions/checkout@v4
+    - name: Set up Python
+      uses: actions/setup-python@v5
       with:
-        pre-build-command: |
-          apt-get --allow-releaseinfo-change update -y && apt-get install -y git && git --version
+        python-version: 3.11
+    - name: Install dependencies
+      run: |
         python -m pip install --upgrade pip setuptools wheel "jax[cpu]"
-          python -m pip install -e .[composable,neural,docs]
-        docs-folder: "docs/"
+        python -m pip install "qunfold @ git+https://github.com/mirkobunse/qunfold@v0.1.4"
+        python -m pip install -e .[neural,docs]
+    - name: Build documentation
+      run: sphinx-build -M html docs/source docs/build
     - name: Publish documentation
       run: |
         git clone ${{ github.server_url }}/${{ github.repository }}.git --branch gh-pages --single-branch __gh-pages/

.github/workflows/pylint.yml

@@ -1,23 +0,0 @@
-name: Pylint
-
-on: [push]
-
-jobs:
-  build:
-    runs-on: ubuntu-latest
-    strategy:
-      matrix:
-        python-version: ["3.8", "3.9", "3.10"]
-    steps:
-    - uses: actions/checkout@v3
-    - name: Set up Python ${{ matrix.python-version }}
-      uses: actions/setup-python@v3
-      with:
-        python-version: ${{ matrix.python-version }}
-    - name: Install dependencies
-      run: |
-        python -m pip install --upgrade pip
-        pip install pylint
-    - name: Analysing the code with pylint
-      run: |
-        pylint $(git ls-files '*.py')

KDEy/kdey_devel.py

@@ -326,7 +326,7 @@ class KDEyMLauto2(KDEyML):
                 if self.target == 'likelihood':
                     loss_fn = neg_loglikelihood_prev
                 else:
-                    loss_fn = lambda prev_hat: qp.error.from_name(self.target)(prev, prev_hat)
+                    loss_fn = lambda prev_hat: qp.error.from_name(self.target)(prevtrue, prev_hat)
 
                 pred_prev, neglikelihood = optim_minimize(loss_fn, init_prev, return_loss=True)
                 loss_accum += neglikelihood

KDEy/quantification_evaluation.py

@@ -43,7 +43,7 @@ METHODS = [
     ('KDEy-AE',  KDEyMLauto2(newLR(), bandwidth='auto', target='mae', search='grid'), wrap_hyper(logreg_grid)),
     ('KDEy-AE+',  KDEyMLauto2(newLR(), bandwidth='auto', target='mae', search='optim'), wrap_hyper(logreg_grid)),
     ('KDEy-RAE',  KDEyMLauto2(newLR(), bandwidth='auto', target='mrae', search='grid'), wrap_hyper(logreg_grid)),
-    ('KDEy-RAE',  KDEyMLauto2(newLR(), bandwidth='auto', target='mrae', search='optim'), wrap_hyper(logreg_grid)),
+    ('KDEy-RAE+',  KDEyMLauto2(newLR(), bandwidth='auto', target='mrae', search='optim'), wrap_hyper(logreg_grid)),
 ]
 
 

KDEy/quantification_evaluation_debug.py

@@ -28,7 +28,7 @@ def plot(xaxis, metrics_measurements, metrics_names, suffix):
     fig, ax1 = plt.subplots(figsize=(8, 6))
 
     def add_plot(ax, mean_error, std_error, name, color, marker):
-        ax.plot(xaxis, mean_error, label=name, marker=marker, color=color)
+        ax.plot(xaxis, mean_error, label=name, marker=marker, color=color, markersize=3)
         if std_error is not None:
             ax.fill_between(xaxis, mean_error - std_error, mean_error + std_error, color=color, alpha=0.2)
 
@@ -74,6 +74,56 @@ def plot(xaxis, metrics_measurements, metrics_names, suffix):
     plt.close()
 
 
+def plot_stack(xaxis, metrics_measurements, metrics_names, suffix):
+
+    # Crear la figura y los ejes (4 bloques verticales)
+    fig, axs = plt.subplots(4, 1, figsize=(8, 12))
+
+    x = xaxis
+
+    indexes = np.arange(len(metrics_measurements))
+    axs_idx = 0
+    # colors = ['b', 'g', 'r', 'c', 'purple']
+    for m_te, m_tr in zip(indexes[:-1:2], indexes[1::2]):
+        metric_te, metric_tr = metrics_measurements[m_te], metrics_measurements[m_tr]
+        metric_te_name, metric_tr_name = metrics_names[m_te], metrics_names[m_tr]
+
+        metric_mean_tr = np.mean(metric_tr, axis=0)
+        metric_std_tr = np.std(metric_tr, axis=0)
+        metric_mean_te = np.mean(metric_te, axis=0)
+        metric_std_te = np.std(metric_te, axis=0)
+
+        axs[axs_idx].plot(xaxis, metric_mean_tr, label=metric_tr_name, marker='o', color='r', markersize=3)
+        axs[axs_idx].fill_between(xaxis, metric_mean_tr - metric_std_tr, metric_mean_tr + metric_std_tr, color='r', alpha=0.2)
+        minx = np.argmin(metric_mean_tr)
+        axs[axs_idx].axvline(xaxis[minx], color='r', linestyle='--', linewidth=1)
+
+        axs[axs_idx].plot(xaxis, metric_mean_te, label=metric_te_name, marker='o', color='b', markersize=3)
+        axs[axs_idx].fill_between(xaxis, metric_mean_te - metric_std_te, metric_mean_te + metric_std_te, color='b', alpha=0.2)
+        minx = np.argmin(metric_mean_te)
+        axs[axs_idx].axvline(xaxis[minx], color='b', linestyle='--', linewidth=1)
+
+        # axs[axs_idx].set_title(f'{metric_te_name} and {metric_tr_name}')
+        axs[axs_idx].legend(loc='lower right')
+        if axs_idx < len(indexes)//2 -1:
+            axs[axs_idx].set_xticks([])
+
+        axs_idx += 1
+
+    # Ajustar el espaciado entre los subplots
+    plt.tight_layout()
+
+    # Mostrar el gráfico
+
+    # Mostrar el gráfico
+    # plt.title(dataset)
+    # plt.show()
+    os.makedirs('./plots/likelihood/', exist_ok=True)
+
+    plt.savefig(f'./plots/likelihood/{dataset}-fig{suffix}.png')
+    plt.close()
+
+
 def generate_data(from_train=False):
     data = qp.datasets.fetch_UCIMulticlassDataset(dataset)
     n_classes = data.n_classes
@@ -110,7 +160,7 @@ def generate_data(from_train=False):
         likelihood_value = []
 
         # for bandwidth in np.linspace(0.01, 0.2, 50):
-        for bandwidth in np.logspace(-5, np.log10(0.2), 50):
+        for bandwidth in np.logspace(-4, np.log10(0.2), 50):
             mix_densities = kde.get_mixture_components(tr_posteriors, tr_y, classes, bandwidth)
             test_densities = [kde.pdf(kde_i, te_posteriors) for kde_i in mix_densities]
 
@@ -172,16 +222,24 @@ for i, dataset in enumerate(tqdm(DATASETS, desc='processing datasets', total=len
     measurement_names = []
     if show_ae:
         measurements.append(AE_error_te)
-        measurement_names.append('AE')
+        measurement_names.append('AE(te)')
+        measurements.append(AE_error_tr)
+        measurement_names.append('AE(tr)')
     if show_rae:
         measurements.append(RAE_error_te)
-        measurement_names.append('RAE')
+        measurement_names.append('RAE(te)')
+        measurements.append(RAE_error_tr)
+        measurement_names.append('RAE(tr)')
     if show_kld:
         measurements.append(KLD_error_te)
-        measurement_names.append('KLD')
+        measurement_names.append('KLD(te)')
+        measurements.append(KLD_error_tr)
+        measurement_names.append('KLD(tr)')
     if show_mse:
         measurements.append(MSE_error_te)
-        measurement_names.append('MSE')
+        measurement_names.append('MSE(te)')
+        measurements.append(MSE_error_tr)
+        measurement_names.append('MSE(tr)')
     measurements.append(normalize_metric(LIKE_value_te))
     measurements.append(normalize_metric(LIKE_value_tr))
     measurement_names.append('NLL(te)')
@@ -200,7 +258,8 @@ for i, dataset in enumerate(tqdm(DATASETS, desc='processing datasets', total=len
     # measurements.append(normalize_metric(LIKE_value_tr))
     # measurement_names.append('NLL(te)')
     # measurement_names.append('NLL(tr)')
-    plot(xaxis, measurements, measurement_names, suffix='AVEtr')
+    # plot(xaxis, measurements, measurement_names, suffix='AVEtr')
+    plot_stack(xaxis, measurements, measurement_names, suffix='AVEtr')
 
 
 

LocalStack/experiments.py

@@ -0,0 +1,126 @@
+import os
+from time import time
+import numpy as np
+from sklearn.linear_model import LogisticRegression
+import quapy as qp
+#from KDEy.kdey_devel import KDEyMLauto, KDEyMLauto2, KDEyMLred
+from LocalStack.method import LocalStackingQuantification, LocalStackingQuantification2
+from quapy.method.aggregative import PACC, EMQ, KDEyML
+from quapy.model_selection import GridSearchQ
+from quapy.protocol import UPP
+from pathlib import Path
+
+SEED = 1
+
+
+
+METHODS = [
+    ('PACC', PACC(), {}),
+    ('EMQ', EMQ(), {}),
+    ('KDEy-ML',  KDEyML(), {}),
+]
+
+TRANSDUCTIVE_METHODS = [
+    ('LSQ',  LocalStackingQuantification(EMQ()), {}),
+    ('LSQ2',  LocalStackingQuantification2(EMQ()), {})
+]
+
+def show_results(result_path):
+    import pandas as pd
+    df = pd.read_csv(result_path + '.csv', sep='\t')
+    pd.set_option('display.max_columns', None)
+    pd.set_option('display.max_rows', None)
+    pd.set_option('display.width', 1000)  # Ajustar el ancho máximo
+    pv = df.pivot_table(index='Dataset', columns="Method", values=["MAE"], margins=True)
+    print(pv)
+    pv = df.pivot_table(index='Dataset', columns="Method", values=["MRAE"], margins=True)
+    print(pv)
+    pv = df.pivot_table(index='Dataset', columns="Method", values=["KLD"], margins=True)
+    print(pv)
+    pv = df.pivot_table(index='Dataset', columns="Method", values=["TR-TIME"], margins=True)
+    print(pv)
+    pv = df.pivot_table(index='Dataset', columns="Method", values=["TE-TIME"], margins=True)
+    print(pv)
+
+
+if __name__ == '__main__':
+
+    qp.environ['SAMPLE_SIZE'] = 500
+    qp.environ['N_JOBS'] = -1
+    n_bags_val = 25
+    n_bags_test = 100
+    result_dir = f'results_quantification/localstack'
+
+    os.makedirs(result_dir, exist_ok=True)
+
+    global_result_path = f'{result_dir}/allmethods'
+    with open(global_result_path + '.csv', 'wt') as csv:
+        csv.write(f'Method\tDataset\tMAE\tMRAE\tKLD\tTR-TIME\tTE-TIME\n')
+
+    for method_name, quantifier, param_grid in METHODS + TRANSDUCTIVE_METHODS:
+
+        print('Init method', method_name)
+
+        with open(global_result_path + '.csv', 'at') as csv:
+            for dataset in qp.datasets.UCI_MULTICLASS_DATASETS:
+                print('init', dataset)
+
+                # run_experiment(global_result_path, method_name, quantifier, param_grid, dataset)
+                local_result_path = os.path.join(Path(global_result_path).parent, method_name + '_' + dataset + '.dataframe')
+
+                if os.path.exists(local_result_path):
+                    print(f'result file {local_result_path} already exist; skipping')
+                    report = qp.util.load_report(local_result_path)
+
+                else:
+                    with qp.util.temp_seed(SEED):
+
+                        data = qp.datasets.fetch_UCIMulticlassDataset(dataset, verbose=True)
+                        train, test = data.train_test
+
+                        transductive_names = [name for (name, *_) in TRANSDUCTIVE_METHODS]
+
+                        if method_name not in transductive_names:
+                            if len(param_grid) == 0:
+                                t_init = time()
+                                quantifier.fit(train)
+                                train_time = time() - t_init
+                            else:
+                                # model selection (train)
+                                train, val = train.split_stratified(random_state=SEED)
+                                protocol = UPP(val, repeats=n_bags_val)
+                                modsel = GridSearchQ(
+                                    quantifier, param_grid, protocol, refit=True, n_jobs=-1, verbose=1, error='mae'
+                                )
+                                t_init = time()
+                                try:
+                                    modsel.fit(train)
+                                    print(f'best params {modsel.best_params_}')
+                                    print(f'best score {modsel.best_score_}')
+                                    quantifier = modsel.best_model()
+                                except:
+                                    print('something went wrong... trying to fit the default model')
+                                    quantifier.fit(train)
+                                train_time = time() - t_init
+                        else:
+                            # transductive
+                            t_init = time()
+                            quantifier.fit(train)  # <-- nothing actually (proyects the X into posteriors only)
+                            train_time = time() - t_init
+
+                        # test
+                        t_init = time()
+                        protocol = UPP(test, repeats=n_bags_test)
+                        report = qp.evaluation.evaluation_report(
+                            quantifier, protocol, error_metrics=['mae', 'mrae', 'kld'], verbose=True
+                        )
+                        test_time = time() - t_init
+                        report['tr_time'] = train_time
+                        report['te_time'] = test_time
+                        report.to_csv(local_result_path)
+
+                means = report.mean(numeric_only=True)
+                csv.write(f'{method_name}\t{dataset}\t{means["mae"]:.5f}\t{means["mrae"]:.5f}\t{means["kld"]:.5f}\t{means["tr_time"]:.3f}\t{means["te_time"]:.3f}\n')
+                csv.flush()
+
+    show_results(global_result_path)

LocalStack/method.py

@@ -0,0 +1,112 @@
+import numpy as np
+import quapy as qp
+from sklearn.multioutput import MultiOutputRegressor
+from sklearn.svm import SVR
+
+from quapy.data import LabelledCollection
+from quapy.method.base import BaseQuantifier
+from quapy.method.aggregative import AggregativeSoftQuantifier
+
+
+class LocalStackingQuantification(BaseQuantifier):
+
+    def __init__(self, surrogate_quantifier, n_samples_gen=200, n_samples_sel=50, comparison_measure='ae', random_state=None):
+        assert isinstance(surrogate_quantifier, AggregativeSoftQuantifier), \
+            f'the surrogate quantifier must be of type {AggregativeSoftQuantifier.__class__.__name__}'
+        self.surrogate_quantifier = surrogate_quantifier
+        self.n_samples_gen = n_samples_gen
+        self.n_samples_sel = n_samples_sel
+        self.comparison_measure = qp.error.from_name(comparison_measure)
+        self.random_state = random_state
+
+    def fit(self, data: LabelledCollection):
+        train, val = data.split_stratified()
+        self.surrogate_quantifier.fit(train)
+        self.val_data = val
+        return self
+
+    def normalize(self, out_simplex:np.ndarray):
+        in_simplex = out_simplex/out_simplex.sum()
+        return in_simplex
+
+    def quantify(self, instances: np.ndarray):
+        assert hasattr(self, 'val_data'), 'quantify called before fit'
+        pred_prevs = self.surrogate_quantifier.quantify(instances)
+        test_size = instances.shape[0]
+
+        samples = []
+        samples_pred_prevs = []
+        samples_distance = []
+        for i in range(self.n_samples_gen):
+            sample_i = self.val_data.sampling(test_size, *pred_prevs, random_state=self.random_state)
+            pred_prev_sample_i = self.surrogate_quantifier.quantify(sample_i.X)
+            err_dist = self.comparison_measure(pred_prevs, pred_prev_sample_i)
+
+            samples.append(sample_i)
+            samples_pred_prevs.append(pred_prev_sample_i)
+            samples_distance.append(err_dist)
+
+        ord_distances = np.argsort(samples_distance)
+        samples_sel = np.asarray(samples)[ord_distances][:self.n_samples_sel]
+        samples_pred_prevs_sel = np.asarray(samples_pred_prevs)[ord_distances][:self.n_samples_sel]
+
+        reg = MultiOutputRegressor(SVR())
+        reg_X = samples_pred_prevs_sel
+        reg_y = [s.prevalence() for s in samples_sel]
+        reg.fit(reg_X, reg_y)
+
+        corrected_prev = reg.predict([pred_prevs])[0]
+
+        corrected_prev = self.normalize(corrected_prev)
+        return corrected_prev
+
+
+
+class LocalStackingQuantification2(BaseQuantifier):
+
+    """
+    Este en vez de seleccionar samples de training para los que la prevalencia predicha se parece a la prevalencia
+    predica en test, saca directamente samples de training con la prevalencia predicha en test
+    """
+
+    def __init__(self, surrogate_quantifier, n_samples_gen=200, n_samples_sel=50, comparison_measure='ae', random_state=None):
+        assert isinstance(surrogate_quantifier, AggregativeSoftQuantifier), \
+            f'the surrogate quantifier must be of type {AggregativeSoftQuantifier.__class__.__name__}'
+        self.surrogate_quantifier = surrogate_quantifier
+        self.n_samples_gen = n_samples_gen
+        self.n_samples_sel = n_samples_sel
+        self.comparison_measure = qp.error.from_name(comparison_measure)
+        self.random_state = random_state
+
+    def fit(self, data: LabelledCollection):
+        train, val = data.split_stratified()
+        self.surrogate_quantifier.fit(train)
+        self.val_data = val
+        return self
+
+    def normalize(self, out_simplex:np.ndarray):
+        in_simplex = out_simplex/out_simplex.sum()
+        return in_simplex
+
+    def quantify(self, instances: np.ndarray):
+        assert hasattr(self, 'val_data'), 'quantify called before fit'
+        pred_prevs = self.surrogate_quantifier.quantify(instances)
+        test_size = instances.shape[0]
+
+        samples = []
+        samples_pred_prevs = []
+        for i in range(self.n_samples_gen):
+            sample_i = self.val_data.sampling(test_size, *pred_prevs, random_state=self.random_state)
+            pred_prev_sample_i = self.surrogate_quantifier.quantify(sample_i.X)
+            samples.append(sample_i)
+            samples_pred_prevs.append(pred_prev_sample_i)
+
+        reg = MultiOutputRegressor(SVR())
+        reg_X = samples_pred_prevs
+        reg_y = [s.prevalence() for s in samples]
+        reg.fit(reg_X, reg_y)
+
+        corrected_prev = reg.predict([pred_prevs])[0]
+
+        corrected_prev = self.normalize(corrected_prev)
+        return corrected_prev

docs/source/conf.py

@@ -11,9 +11,14 @@ import sys
 from os.path import join
 quapy_path = join(pathlib.Path(__file__).parents[2].resolve().as_posix(), 'quapy')
 wiki_path = join(pathlib.Path(__file__).parents[0].resolve().as_posix(), 'wiki')
+source_path = pathlib.Path(__file__).parents[2].resolve().as_posix()
 print(f'quapy path={quapy_path}')
+print(f'quapy source path={source_path}')
 sys.path.insert(0, quapy_path)
 sys.path.insert(0, wiki_path)
+sys.path.insert(0, source_path)
+
+print(sys.path)
 
 
 project = 'QuaPy: A Python-based open-source framework for quantification'

docs/source/manuals/methods.md

@@ -447,7 +447,7 @@ The [](quapy.method.composable) module allows the composition of quantification
 ```sh
 pip install --upgrade pip setuptools wheel
 pip install "jax[cpu]"
-pip install quapy[composable]
+pip install "qunfold @ git+https://github.com/mirkobunse/qunfold@v0.1.4"
 ```
 
 ### Basics

examples/14.composable_methods.py

@@ -2,6 +2,13 @@
 This example illustrates the composition of quantification methods from
 arbitrary loss functions and feature transformations. It will extend the basic
 example on the usage of quapy with this composition.
+
+This example requires the installation of qunfold, the back-end of QuaPy's
+composition module:
+
+    pip install --upgrade pip setuptools wheel
+    pip install "jax[cpu]"
+    pip install "qunfold @ git+https://github.com/mirkobunse/qunfold@v0.1.4"
 """
 
 import numpy as np

quapy/method/composable.py

@@ -1,5 +1,15 @@
 """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."""
 
+_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@v0.1.4"
+"""
+
+try:
     import qunfold
     from qunfold.quapy import QuaPyWrapper
     from qunfold.sklearn import CVClassifier
@@ -40,6 +50,8 @@ __all__ = [ # control public members, e.g., for auto-documentation in sphinx; om
         "GaussianKernelTransformer",
         "GaussianRFFKernelTransformer",
     ]
+except ImportError as e:
+    raise ImportError(_import_error_message) from e
 
 def ComposableQuantifier(loss, transformer, **kwargs):
     """A generic quantification / unfolding method that solves a linear system of equations.

setup.py

@@ -125,7 +125,6 @@ setup(
     # projects.
     extras_require={  # Optional
        'bayes': ['jax', 'jaxlib', 'numpyro'],
-       'composable': ['qunfold @ git+https://github.com/mirkobunse/qunfold@v0.1.4'],
        'neural': ['torch'],
        'tests': ['certifi'],
        'docs' : ['sphinx-rtd-theme', 'myst-parser'],