forecastutils 0.1.0__tar.gz

forecastutils-0.1.0/PKG-INFO

@@ -0,0 +1,15 @@
+Metadata-Version: 2.4
+Name: forecastutils
+Version: 0.1.0
+Summary: Custom utilities for prediction pipeline
+Requires-Dist: pytest
+Requires-Dist: scikit-learn
+Requires-Dist: xgboost
+Requires-Dist: pandas
+Requires-Dist: numpy
+Requires-Dist: holidays
+Requires-Dist: vacances_scolaires_france
+Requires-Dist: matplotlib
+Requires-Dist: requests
+Dynamic: requires-dist
+Dynamic: summary

forecastutils-0.1.0/forecastutils.egg-info/PKG-INFO

@@ -0,0 +1,15 @@
+Metadata-Version: 2.4
+Name: forecastutils
+Version: 0.1.0
+Summary: Custom utilities for prediction pipeline
+Requires-Dist: pytest
+Requires-Dist: scikit-learn
+Requires-Dist: xgboost
+Requires-Dist: pandas
+Requires-Dist: numpy
+Requires-Dist: holidays
+Requires-Dist: vacances_scolaires_france
+Requires-Dist: matplotlib
+Requires-Dist: requests
+Dynamic: requires-dist
+Dynamic: summary

forecastutils-0.1.0/forecastutils.egg-info/SOURCES.txt

@@ -0,0 +1,8 @@
+setup.py
+forecastutils.egg-info/PKG-INFO
+forecastutils.egg-info/SOURCES.txt
+forecastutils.egg-info/dependency_links.txt
+forecastutils.egg-info/requires.txt
+forecastutils.egg-info/top_level.txt
+test/test_core.py
+test/test_preprocessing.py

forecastutils-0.1.0/forecastutils.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

forecastutils-0.1.0/forecastutils.egg-info/requires.txt

@@ -0,0 +1,9 @@
+pytest
+scikit-learn
+xgboost
+pandas
+numpy
+holidays
+vacances_scolaires_france
+matplotlib
+requests

forecastutils-0.1.0/forecastutils.egg-info/top_level.txt

@@ -0,0 +1 @@
+

forecastutils-0.1.0/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

forecastutils-0.1.0/setup.py

@@ -0,0 +1,17 @@
+from setuptools import setup, find_packages
+import pathlib
+
+# Lit le requirements.txt
+this_dir = pathlib.Path(__file__).parent
+requirements_path = this_dir / "requirements.txt"
+with open(requirements_path, "r") as f:
+    requirements = f.read().splitlines()
+
+setup(
+    name="forecastutils",
+    version="0.1.0",
+    packages=find_packages(),
+    install_requires=requirements,
+    include_package_data=True,
+    description="Custom utilities for prediction pipeline",
+)

forecastutils-0.1.0/test/test_core.py

@@ -0,0 +1,43 @@
+import pytest
+import pandas as pd
+import numpy as np
+from sklearn.pipeline import Pipeline
+from sklearn.linear_model import Ridge
+from sklearn.preprocessing import StandardScaler
+import forecastutils.core as fcore
+
+def test_build_model_forecast():
+    # Données synthétiques
+    n_samples = 100
+    X = pd.DataFrame({'feature': np.arange(n_samples)})
+    y = np.arange(n_samples) + np.random.normal(0, 1, n_samples)
+
+    # Pipeline simple
+    pipeline = Pipeline([
+        ('scaler', StandardScaler()),
+        ('regressor', Ridge())
+    ])
+
+    # Grille de paramètres
+    param_grids = {
+        'pipeline_0': {
+            'regressor__alpha': [0.1, 1.0, 10.0]
+        }
+    }
+
+    # Appel de la fonction
+    results = fcore.build_model_forecast(
+        pipelines=[pipeline],
+        param_grids=param_grids,
+        X=X,
+        y=y,
+        scoring='neg_mean_absolute_error',
+        n_splits=5,
+        verbose=0
+    )
+
+    # Vérifications
+    assert 'pipeline_0' in results, "Le nom du pipeline n'est pas dans les résultats"
+    assert 'best_model' in results['pipeline_0'], "Le modèle optimal n'est pas présent"
+    assert isinstance(results['pipeline_0']['best_score_cv'], float), "Le score CV doit être un float"
+    assert isinstance(results['pipeline_0']['best_params'], dict), "Les paramètres optimaux doivent être un dictionnaire"

forecastutils-0.1.0/test/test_preprocessing.py

@@ -0,0 +1,101 @@
+import pytest
+import pandas as pd
+import numpy as np
+import forecastutils.preprocessing as pp
+
+def test_month_binary_encoder():
+    # Données de test
+    df = pd.DataFrame({
+        'date': ['2023-01-15', '2023-06-10', '2023-12-25']
+    })
+
+    # Transformer
+    encoder = pp.MonthBinaryEncoder(date_column='date')
+    transformed_df = encoder.fit_transform(df)
+
+    # Vérifications
+    assert 'is_janvier' in transformed_df.columns
+    assert 'is_mai' in transformed_df.columns
+    assert 'is_decembre' in transformed_df.columns
+    assert transformed_df.loc[0, 'is_janvier'] == True
+    assert transformed_df.loc[0, 'is_decembre'] == False
+    assert transformed_df.loc[1, 'is_juin'] == True
+    assert transformed_df.loc[2, 'is_decembre'] == True
+
+    # Vérifie qu’il n’y a qu’un seul mois actif par ligne
+    for i in range(len(transformed_df)):
+        mois_actifs = [col for col in transformed_df.columns if col.startswith('is_') and transformed_df.loc[i, col]]
+        assert len(mois_actifs) == 1
+
+
+def test_monthly_sum_lag_transformer():
+    # Données de test
+    data = pd.DataFrame({
+        'date': pd.date_range(start='2023-01-01', periods=60, freq='D'),
+        'value': np.ones(60)
+    })
+
+    # Transformer avec lag de 1 mois
+    transformer = pp.MonthlySumLagTransformer(date_col='date', value_col='value', lags=[1], fillna=False)
+    transformed = transformer.transform(data)
+
+    # Vérifications
+    assert 'value_monthly_sum_lag1' in transformed.columns
+    assert 'month' not in transformed.columns
+    assert 'value_monthly_sum' not in transformed.columns
+
+    # Vérifie que les valeurs de lag sont correctes
+    jan_sum = data[data['date'].dt.month == 1]['value'].sum()
+    feb_sum = data[data['date'].dt.month == 2]['value'].sum()
+
+    # Les lignes de février doivent avoir la somme de janvier
+    feb_dates = transformed[transformed['date'].dt.month == 2]
+    assert all(feb_dates['value_monthly_sum_lag1'] == jan_sum)
+
+    # Les lignes de janvier doivent etre nulles 
+    jan_dates = transformed[transformed['date'].dt.month == 1]
+    assert jan_dates['value_monthly_sum_lag1'].isna().all()
+
+
+def test_moving_average_delta_transformer():
+    # Données de test
+    data = pd.DataFrame({
+        'date': pd.date_range(start='2023-01-01', periods=30, freq='D'),
+        'value': range(30)
+    })
+
+    # Initialisation du transformer
+    transformer = pp.MovingAverageDeltaTransformer(date_column='date', value_column='value', window_pairs=[(3, 5)])
+
+    # Transformation
+    transformed = transformer.transform(data)
+
+    # Vérifications
+    assert 'delta_ma_3_5' in transformed.columns, "La colonne delta n'a pas été ajoutée"
+    assert len(transformed) == len(data), "La taille du DataFrame transformé est incorrecte"
+    assert transformed['delta_ma_3_5'][3] == 0.5, "Le calcul du delta de moving average est faux"
+    assert not transformed['delta_ma_3_5'].isnull().all(), "Toutes les valeurs delta sont nulles"
+
+
+def test_remaining_weekdays_excluding_today():
+    df = pd.DataFrame({
+        'date': pd.to_datetime(['2023-10-25']),  # Mercredi
+        'value': [42]
+    })
+
+    transformer = pp.RemainingWeekdaysInMonthTransformer(date_column='date')
+    transformed_df = transformer.transform(df)
+
+    expected_counts = {
+        'Monday': 1,     # 30
+        'Tuesday': 1,    # 31
+        'Wednesday': 0,  # 25 est exclu, pas d'autre mercredi
+        'Thursday': 1,   # 26
+        'Friday': 1,     # 27
+        'Saturday': 1,   # 28
+        'Sunday': 1      # 29
+    }
+
+    for day, expected in expected_counts.items():
+        actual = transformed_df.loc[0, day]
+        assert actual == expected, f"Erreur pour {day}: attendu {expected}, obtenu {actual}"