ADD: Model tests

choice_learn/datasets/base.py

@@ -4,6 +4,7 @@
 import gzip
 import os
 from importlib import resources
+from pathlib import Path
 
 import numpy as np
 import pandas as pd
@@ -12,7 +13,7 @@
 from choice_learn.data.choice_dataset import ChoiceDataset
 
 OS_DATA_MODULE = os.path.join(os.path.abspath(".."), "choice_learn", "datasets", "data")
-DATA_MODULE = "choice_learn.datasets.data"
+DATA_MODULE = "choice_learn/datasets/data"
 
 
 def get_path(data_file_name, module=DATA_MODULE):
@@ -35,10 +36,12 @@ def get_path(data_file_name, module=DATA_MODULE):
     import sys
 
     if sys.version >= "3.9":
-        return resources.files(module) / data_file_name
+        return resources.files(module.replace("/", ".")) / data_file_name
 
-    with resources.path(module, data_file_name) as path:
-        return path
+    # with resources.path(module, data_file_name) as path:
+    #     return path
+    path = Path(module).resolve() / data_file_name
+    return path.as_posix()
 
 
 def load_csv(data_file_name, data_module=DATA_MODULE, encoding="utf-8"):
@@ -140,7 +143,7 @@ def download_from_url(url):
     full_path = get_path(local_filename, module=DATA_MODULE)
 
     # Check that the file is not already downloaded in the DATA_MODULE directory
-    if not os.path.isfile(full_path):
+    if not Path.is_file(Path(full_path)):
         print(f"Downloading {local_filename} from {url}")
         try:
             with requests.get(url, stream=True, timeout=20) as r:
@@ -152,7 +155,7 @@ def download_from_url(url):
             print(f"Couldn't download automatically the dataset from {url}")
 
         # Move the downloaded file to the DATA_MODULE directory
-        os.rename(local_filename, full_path)
+        Path(local_filename).rename(full_path)
         print(f"Download completed. File saved as {local_filename} in {full_path}")
 
     return local_filename

choice_learn/datasets/expedia.py

@@ -10,8 +10,8 @@
 from choice_learn.data.storage import OneHotStorage
 from choice_learn.datasets.base import get_path
 
-DATA_MODULE = "choice_learn.datasets.data"
-CACHE_MODULE = "choice_learn.datasets.cache"
+DATA_MODULE = "choice_learn/datasets/data"
+CACHE_MODULE = "choice_learn/datasets/cache"
 
 
 def load_expedia(as_frame=False, preprocessing="rumnet"):
@@ -26,7 +26,7 @@ def load_expedia(as_frame=False, preprocessing="rumnet"):
     """
     filename = "expedia.csv"
     data_path = get_path(filename, module=DATA_MODULE)
-    if not Path.exists(data_path):
+    if not Path.exists(Path(data_path)):
         print("In order to use the Expedia dataset, please download it from:")
         print("https://www.kaggle.com/c/expedia-personalized-sort")
         print("and save it in the following location:")

choice_learn/datasets/tafeng.py

@@ -6,7 +6,7 @@
 from choice_learn.data.choice_dataset import ChoiceDataset
 from choice_learn.datasets.base import get_path
 
-DATA_MODULE = "choice_learn.datasets.data"
+DATA_MODULE = "choice_learn/datasets/data"
 
 
 def load_tafeng(as_frame=False, return_desc=False, preprocessing=None):

tests/integration_tests/models/test_reslogit.py

@@ -1,12 +1,14 @@
 """Tests ResLogit."""
 
 import numpy as np
+import pytest
 import tensorflow as tf
 
 from choice_learn.datasets import load_swissmetro
 
 # from choice_learn.models import ResLogit, SimpleMNL
 from choice_learn.models import ResLogit
+from choice_learn.models.reslogit import ResLayer
 
 dataset = load_swissmetro()
 dataset = dataset[:100]  # Reduce the dataset size for faster testing
@@ -352,3 +354,12 @@ def test_that_endpoints_run():
     model.evaluate(dataset, mode="optim")
     model.predict_probas(dataset)
     assert True
+
+
+def test_activation():
+    """Tests ResLogit activation."""
+    layer = ResLayer()
+    for act in ["linear", "relu", "-relu", "softplus", "tanh", "sigmoid"]:
+        _ = layer.get_activation_function(act)
+    with pytest.raises(ValueError):
+        layer.get_activation_function("abc")

tests/unit_tests/datasets/test_download.py

@@ -0,0 +1,9 @@
+"""Test the file downloading function."""
+
+from choice_learn.datasets.base import download_from_url
+
+
+def test_download():
+    """Tests downloading a dummy csv file."""
+    url = "https://github.com/artefactory/choice-learn/blob/main/tests/data/test_data.csv"
+    download_from_url(url)

tests/unit_tests/models/test_tastenet.py

@@ -0,0 +1,146 @@
+"""Simple tests for the TasteNet model."""
+
+import pytest
+
+from choice_learn.datasets import load_swissmetro
+from choice_learn.models import TasteNet
+
+customers_id, dataset = load_swissmetro(preprocessing="tastenet", as_frame=False)
+dataset = dataset[:20]
+
+taste_net_layers = []
+taste_net_activation = "relu"
+items_features_by_choice_parametrization = [
+    [-1.0, "-exp", "-exp", 0.0, "linear", 0.0, 0.0],
+    [-1.0, "-exp", "-exp", "linear", 0.0, "linear", 0.0],
+    [-1.0, "-exp", 0.0, 0.0, 0.0, 0.0, 0.0],
+]
+
+
+def test_activation():
+    """Tests TasteNet activation."""
+    tastenet = TasteNet(
+        taste_net_layers=taste_net_layers,
+        taste_net_activation=taste_net_activation,
+        items_features_by_choice_parametrization=items_features_by_choice_parametrization,
+        optimizer="Adam",
+        epochs=40,
+        lr=0.001,
+        batch_size=32,
+    )
+    for act in ["linear", "relu", "-relu", "exp", "-exp", "tanh", "sigmoid"]:
+        _ = tastenet.get_activation_function(act)
+    with pytest.raises(ValueError):
+        tastenet.get_activation_function("abc")
+
+
+def test_fit_adam_and_predict():
+    """Test fit with Gradient Descent."""
+    tastenet = TasteNet(
+        taste_net_layers=taste_net_layers,
+        taste_net_activation=taste_net_activation,
+        items_features_by_choice_parametrization=items_features_by_choice_parametrization,
+        optimizer="Adam",
+        epochs=5,
+        lr=0.001,
+        batch_size=32,
+    )
+    _ = tastenet.fit(dataset)
+    _ = tastenet.predict_tastes(dataset.shared_features_by_choice)
+    assert True
+
+
+def test_fit_lbfgs():
+    """Test fit with Gradient Descent."""
+    tastenet = TasteNet(
+        taste_net_layers=[4],
+        taste_net_activation=taste_net_activation,
+        items_features_by_choice_parametrization=items_features_by_choice_parametrization,
+        optimizer="lbfgs",
+        epochs=5,
+        lr=0.001,
+        batch_size=32,
+    )
+    _ = tastenet.fit(dataset)
+    assert True
+
+
+def test_errors_raised():
+    """Test diverse errors that should be raised."""
+    with pytest.raises(ValueError):
+        tastenet = TasteNet(
+            taste_net_layers=taste_net_layers,
+            taste_net_activation=taste_net_activation,
+            items_features_by_choice_parametrization=[
+                [-1.0, "-exp", "-exp", 0.0, "linear", 0.0, 0.0],
+                [-1.0, "-exp", "-exp", "linear", 0.0, "linear", 0.0],
+            ],
+            optimizer="Adam",
+            epochs=5,
+            lr=0.001,
+            batch_size=32,
+        )
+        _ = tastenet.fit(dataset)
+
+    with pytest.raises(ValueError):
+        tastenet = TasteNet(
+            taste_net_layers=taste_net_layers,
+            taste_net_activation=taste_net_activation,
+            items_features_by_choice_parametrization=[
+                [-1.0, "-exp", "-exp", 0.0, "linear", 0.0, 0.0, 0.0],
+                [-1.0, "-exp", "-exp", "linear", 0.0, "linear", 0.0, 0.0],
+                [-1.0, "-exp", 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],
+            ],
+            optimizer="Adam",
+            epochs=5,
+            lr=0.001,
+            batch_size=32,
+        )
+        _ = tastenet.fit(dataset)
+
+    with pytest.raises(ValueError):
+        tastenet = TasteNet(
+            taste_net_layers=taste_net_layers,
+            taste_net_activation=taste_net_activation,
+            items_features_by_choice_parametrization=[
+                [-1.0, "-exp", "-exp", 0.0, "linear", 0.0, 0.0],
+                [-1.0, "-exp", "-exp", "linear", 0.0, "linear", 0.0],
+            ],
+            optimizer="lbfgs",
+            epochs=5,
+            lr=0.001,
+            batch_size=32,
+        )
+        _ = tastenet.fit(dataset)
+
+    with pytest.raises(ValueError):
+        tastenet = TasteNet(
+            taste_net_layers=taste_net_layers,
+            taste_net_activation=taste_net_activation,
+            items_features_by_choice_parametrization=[
+                [-1.0, "-exp", "-exp", 0.0, "linear", 0.0, 0.0, 0.0],
+                [-1.0, "-exp", "-exp", "linear", 0.0, "linear", 0.0, 0.0],
+                [-1.0, "-exp", 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],
+            ],
+            optimizer="lbfgs",
+            epochs=5,
+            lr=0.001,
+            batch_size=32,
+        )
+        _ = tastenet.fit(dataset)
+
+    with pytest.raises(ValueError):
+        tastenet = TasteNet(
+            taste_net_layers=taste_net_layers,
+            taste_net_activation=taste_net_activation,
+            items_features_by_choice_parametrization=[
+                [-1.0, "-exp", "-exp", 0.0, "linear", 0.0, 0.0],
+                [-1.0, "-exp", "-exp", "linear", 0.0, "linear", 0.0, 0.0],
+                [-1.0, "-exp", 0.0, 0.0, 0.0, 0.0, 0.0],
+            ],
+            optimizer="lbfgs",
+            epochs=5,
+            lr=0.001,
+            batch_size=32,
+        )
+        _ = tastenet.fit(dataset)