✅ Test sklearn transformer interfaces

tests/models/test_transformers.py

@@ -0,0 +1,48 @@
+"""Test scikit-learn transformers provided by PIMMS."""
+import numpy as np
+import pandas as pd
+import pytest
+
+from pimmslearn.sklearn.ae_transformer import AETransformer
+from pimmslearn.sklearn.cf_transformer import CollaborativeFilteringTransformer
+
+test_data = 'project/data/dev_datasets/HeLa_6070/protein_groups_wide_N50_M227.csv'
+index_name = 'Sample ID'
+column_name = 'protein group'
+value_name = 'intensity'
+
+
+def test_CollaborativeFilteringTransformer():
+    model = CollaborativeFilteringTransformer(
+        target_column=value_name,
+        sample_column=index_name,
+        item_column=column_name,)
+    # read data, name index and columns
+    df = pd.read_csv(test_data, index_col=0)
+    df = np.log2(df + 1)
+    df.index.name = index_name  # already set
+    df.columns.name = column_name  # not set due to csv disk file format
+    series = df.stack()
+    series.name = value_name  # ! important
+    # run for 2 epochs
+    model.fit(series, cuda=False, epochs_max=2)
+
+
+@pytest.mark.parametrize("model", ['DAE', 'VAE'])
+def test_AETransformer(model):
+    df = pd.read_csv(test_data, index_col=0)
+    df = np.log2(df + 1)
+
+    df.index.name = index_name  # already set
+    df.columns.name = column_name  # not set due to csv disk file format
+    model = AETransformer(
+        model=model,
+        hidden_layers=[512,],
+        latent_dim=50,
+        out_folder='runs/scikit_interface',
+        batch_size=10,
+    )
+    model.fit(df,
+              cuda=False,
+              epochs_max=2,
+              )