more tests and documentation

Author: JohnMount

data_algebra/cdata_impl.py

@@ -132,7 +132,7 @@ def transform(
 
     def compose(self, other):
         """
-        Compose transforms
+        Experimental method to compose transforms
         (self.compose(other)).transform(data) == self.transform(other.transform(data))
 
         :param other: another data_algebra.cdata_impl.RecordMap
@@ -183,7 +183,7 @@ def __rrshift__(self, other):  # override other >> self
         if other is None:
             return self
         if isinstance(other, RecordMap):
-            # data >> other >> self
+            # (data >> other) >> self == data >> (other >> self)
             return self.compose(other)
         return self.transform(other)
 

tests/test_cdata_example.py

@@ -8,6 +8,7 @@
 from data_algebra.cdata_impl import RecordMap
 import data_algebra.yaml
 from data_algebra.cdata_impl import record_map_from_simple_obj
+import data_algebra.util
 
 
 def test_cdata_example():
@@ -50,8 +51,67 @@ def test_cdata_example():
     back = data_algebra.yaml.to_pipeline(yaml_obj)
 
 
-def test_cdata_transport():
+def test_keras_example():
     obj = {'blocks_out': {'record_keys': 'epoch', 'control_table_keys': 'measure',
                     'control_table': {'measure': ['minus binary cross entropy', 'accuracy'],
                                       'training': ['loss', 'acc'], 'validation': ['val_loss', 'val_acc']}}}
-    record_map_from_simple_obj(obj)
+    record_map = record_map_from_simple_obj(obj)
+    data = pandas.DataFrame({
+        'val_loss': [-0.377, -0.2997],
+        'val_acc': [0.8722, 0.8895],
+        'loss': [-0.5067, -0.3002],
+        'acc': [0.7852, 0.904],
+        'epoch': [1, 2],
+    })
+    res = record_map.transform(data)
+    expect = pandas.DataFrame({
+        'epoch': [1, 1, 2, 2],
+        'measure': ['accuracy', 'minus binary cross entropy', 'accuracy', 'minus binary cross entropy'],
+        'training': [0.7852, -0.5067, 0.9040, -0.3002],
+        'validation': [0.8722, -0.3770, 0.8895, -0.2997],
+    })
+    assert data_algebra.util.equivalent_frames(res, expect)
+
+def test_cdata_block():
+    data = pandas.DataFrame({
+        'record_id': [1, 1, 1, 2, 2, 2],
+        'row': ['row1', 'row2', 'row3', 'row1', 'row2', 'row3'],
+        'col1': [1, 4, 7, 11, 14, 17],
+        'col2': [2, 5, 8, 12, 15, 18],
+        'col3': [3, 6, 9, 13, 16, 19],
+    })
+
+    record_keys = ['record_id']
+
+    incoming_shape = pandas.DataFrame({
+        'row': ['row1', 'row2', 'row3'],
+        'col1': ['v11', 'v21', 'v31'],
+        'col2': ['v12', 'v22', 'v32'],
+        'col3': ['v13', 'v23', 'v33'],
+    })
+
+    outgoing_shape = pandas.DataFrame({
+        'column_label': ['rec_col1', 'rec_col2', 'rec_col3'],
+        'c_row1': ['v11', 'v12', 'v13'],
+        'c_row2': ['v21', 'v22', 'v23'],
+        'c_row3': ['v31', 'v32', 'v33'],
+    })
+
+    record_map = data_algebra.cdata_impl.RecordMap(
+        blocks_in=data_algebra.cdata.RecordSpecification(
+            control_table=incoming_shape,
+            record_keys=record_keys
+        ),
+        blocks_out=data_algebra.cdata.RecordSpecification(
+            control_table=outgoing_shape,
+            record_keys=record_keys
+        ),
+    )
+
+    res = record_map.transform(data)
+
+    inv = record_map.inverse()
+
+    back = inv.transform(res)
+
+    assert data_algebra.util.equivalent_frames(data, back)