Update tests to verify Component Graph support in pipelines (#2830)

* add tests

* update release notes

* update docstrings and test

* rename comp graph vars

* fixing docs

docs/source/release_notes.rst

@@ -16,6 +16,7 @@ Release Notes
         * Fixed bug where ``calculate_permutation_importance`` was not calculating the right value for pipelines with target transformers :pr:`2782`
         * Fixed bug where transformed target values were not used in ``fit`` for time series pipelines :pr:`2780`
         * Fixed bug where ``score_pipelines`` method of ``AutoMLSearch`` would not work for time series problems :pr:`2786`
+        * Add tests to verify ``ComponentGraph`` support by pipelines :pr:`2830`
     * Changes
         * Changed woodwork initialization to use partial schemas :pr:`2774`
         * Made ``Transformer.transform()`` an abstract method :pr:`2744`

docs/source/user_guide/pipelines.ipynb

@@ -28,9 +28,11 @@
     " - `TimeSeriesMulticlassClassificationPipeline`\n",
     "    \n",
     "The class you want to use will depend on your problem type.\n",
-    "The only required parameter input for instantiating a pipeline instance is `component_graph`, which is either a list or a dictionary containing a sequence of components to be fit and evaluated.\n",
+    "The only required parameter input for instantiating a pipeline instance is `component_graph`, which can be a `ComponentGraph` [instance](https://evalml.alteryx.com/en/stable/autoapi/evalml/pipelines/index.html#evalml.pipelines.ComponentGraph), a list, or a dictionary containing a sequence of components to be fit and evaluated.\n",
     "\n",
-    "A `component_graph` list is the default representation, which represents a linear order of transforming components with an estimator as the final component. A `component_graph` dictionary is used to represent a non-linear graph of components, where the key is a unique name for each component and the value is a list with the component's class as the first element and any parents of the component as the following element(s). For either `component_graph` format, each component can be provided as a reference to the component class for custom components, and as either a string name or as a reference to the component class for components defined in EvalML."
+    "A `component_graph` list is the default representation, which represents a linear order of transforming components with an estimator as the final component. A `component_graph` dictionary is used to represent a non-linear graph of components, where the key is a unique name for each component and the value is a list with the component's class as the first element and any parents of the component as the following element(s). For these two `component_graph` formats, each component can be provided as a reference to the component class for custom components, and as either a string name or as a reference to the component class for components defined in EvalML.\n",
+    "\n",
+    "If you choose to provide a `ComponentGraph` instance and want to set custom parameters for your pipeline, set it through the pipeline initialization rather than `ComponentGraph.instantiate()`."
    ]
   },
   {
@@ -39,7 +41,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "from evalml.pipelines import MulticlassClassificationPipeline\n",
+    "from evalml.pipelines import MulticlassClassificationPipeline, ComponentGraph\n",
     "\n",
     "component_graph_as_list = ['Imputer', 'Random Forest Classifier']\n",
     "MulticlassClassificationPipeline(component_graph=component_graph_as_list)"
@@ -62,6 +64,18 @@
     "MulticlassClassificationPipeline(component_graph=component_graph_as_dict)"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "cg = ComponentGraph(component_graph_as_dict)\n",
+    "\n",
+    "# set parameters in the pipeline rather than through cg.instantiate()\n",
+    "MulticlassClassificationPipeline(component_graph=cg, parameters={})"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -481,4 +495,4 @@
  },
  "nbformat": 4,
  "nbformat_minor": 4
-}
+}

evalml/pipelines/binary_classification_pipeline.py

@@ -15,7 +15,8 @@ class BinaryClassificationPipeline(
     """Pipeline subclass for all binary classification pipelines.
 
     Args:
-        component_graph (list or dict): List of components in order. Accepts strings or ComponentBase subclasses in the list.
+        component_graph (ComponentGraph, list, dict): ComponentGraph instance, list of components in order, or dictionary of components.
+            Accepts strings or ComponentBase subclasses in the list.
             Note that when duplicate components are specified in a list, the duplicate component names will be modified with the
             component's index in the list. For example, the component graph
             [Imputer, One Hot Encoder, Imputer, Logistic Regression Classifier] will have names

evalml/pipelines/multiclass_classification_pipeline.py

@@ -7,7 +7,8 @@ class MulticlassClassificationPipeline(ClassificationPipeline):
     """Pipeline subclass for all multiclass classification pipelines.
 
     Args:
-        component_graph (list or dict): List of components in order. Accepts strings or ComponentBase subclasses in the list.
+        component_graph (ComponentGraph, list, dict): ComponentGraph instance, list of components in order, or dictionary of components.
+            Accepts strings or ComponentBase subclasses in the list.
             Note that when duplicate components are specified in a list, the duplicate component names will be modified with the
             component's index in the list. For example, the component graph
             [Imputer, One Hot Encoder, Imputer, Logistic Regression Classifier] will have names

evalml/pipelines/pipeline_base.py

@@ -43,7 +43,8 @@ class PipelineBase(ABC, metaclass=PipelineBaseMeta):
     """Machine learning pipeline.
 
     Args:
-        component_graph (list or dict): List of components in order. Accepts strings or ComponentBase subclasses in the list.
+        component_graph (ComponentGraph, list, dict): ComponentGraph instance, list of components in order, or dictionary of components.
+            Accepts strings or ComponentBase subclasses in the list.
             Note that when duplicate components are specified in a list, the duplicate component names will be modified with the
             component's index in the list. For example, the component graph
             [Imputer, One Hot Encoder, Imputer, Logistic Regression Classifier] will have names

evalml/pipelines/regression_pipeline.py

@@ -8,7 +8,8 @@ class RegressionPipeline(PipelineBase):
     """Pipeline subclass for all regression pipelines.
 
     Args:
-        component_graph (list or dict): List of components in order. Accepts strings or ComponentBase subclasses in the list.
+        component_graph (ComponentGraph, list, dict): ComponentGraph instance, list of components in order, or dictionary of components.
+            Accepts strings or ComponentBase subclasses in the list.
             Note that when duplicate components are specified in a list, the duplicate component names will be modified with the
             component's index in the list. For example, the component graph
             [Imputer, One Hot Encoder, Imputer, Logistic Regression Classifier] will have names

evalml/pipelines/time_series_classification_pipelines.py

@@ -16,7 +16,8 @@ class TimeSeriesClassificationPipeline(TimeSeriesPipelineBase, ClassificationPip
     """Pipeline base class for time series classification problems.
 
     Args:
-        component_graph (list or dict): List of components in order. Accepts strings or ComponentBase subclasses in the list.
+        component_graph (ComponentGraph, list, dict): ComponentGraph instance, list of components in order, or dictionary of components.
+            Accepts strings or ComponentBase subclasses in the list.
             Note that when duplicate components are specified in a list, the duplicate component names will be modified with the
             component's index in the list. For example, the component graph
             [Imputer, One Hot Encoder, Imputer, Logistic Regression Classifier] will have names

evalml/pipelines/time_series_pipeline_base.py

@@ -11,7 +11,8 @@ class TimeSeriesPipelineBase(PipelineBase, metaclass=PipelineBaseMeta):
     """Pipeline base class for time series problems.
 
     Args:
-        component_graph (list or dict): List of components in order. Accepts strings or ComponentBase subclasses in the list.
+        component_graph (ComponentGraph, list, dict): ComponentGraph instance, list of components in order, or dictionary of components.
+            Accepts strings or ComponentBase subclasses in the list.
             Note that when duplicate components are specified in a list, the duplicate component names will be modified with the
             component's index in the list. For example, the component graph
             [Imputer, One Hot Encoder, Imputer, Logistic Regression Classifier] will have names

evalml/pipelines/time_series_regression_pipeline.py

@@ -7,7 +7,8 @@ class TimeSeriesRegressionPipeline(TimeSeriesPipelineBase):
     """Pipeline base class for time series regression problems.
 
     Args:
-        component_graph (list or dict): List of components in order. Accepts strings or ComponentBase subclasses in the list.
+        component_graph (ComponentGraph, list, dict): ComponentGraph instance, list of components in order, or dictionary of components.
+            Accepts strings or ComponentBase subclasses in the list.
             Note that when duplicate components are specified in a list, the duplicate component names will be modified with the
             component's index in the list. For example, the component graph
             [Imputer, One Hot Encoder, Imputer, Logistic Regression Classifier] will have names

evalml/tests/pipeline_tests/test_pipelines.py

@@ -2796,3 +2796,81 @@ def test_training_only_component_in_pipeline_transform(X_y_binary):
     pipeline.fit(X, y)
     transformed = pipeline.transform(X)
     assert len(transformed) == len(X) - 2
+
+
+def test_component_graph_pipeline():
+    classification_cg = ComponentGraph(
+        {
+            "Imputer": ["Imputer", "X", "y"],
+            "Undersampler": ["Undersampler", "Imputer.x", "y"],
+            "Logistic Regression Classifier": [
+                "Logistic Regression Classifier",
+                "Undersampler.x",
+                "Undersampler.y",
+            ],
+        }
+    )
+
+    regression_cg = ComponentGraph(
+        {
+            "Imputer": ["Imputer", "X", "y"],
+            "Linear Regressor": [
+                "Linear Regressor",
+                "Imputer.x",
+                "y",
+            ],
+        }
+    )
+
+    no_estimator_cg = ComponentGraph(
+        {
+            "Imputer": ["Imputer", "X", "y"],
+            "Undersampler": ["Undersampler", "Imputer.x", "y"],
+        }
+    )
+
+    assert (
+        BinaryClassificationPipeline(classification_cg).component_graph
+        == classification_cg
+    )
+    assert RegressionPipeline(regression_cg).component_graph == regression_cg
+    assert (
+        BinaryClassificationPipeline(no_estimator_cg).component_graph == no_estimator_cg
+    )
+    with pytest.raises(
+        ValueError, match="Problem type regression not valid for this component graph"
+    ):
+        RegressionPipeline(classification_cg)
+
+
+def test_component_graph_pipeline_initialized():
+    component_graph1 = ComponentGraph(
+        {
+            "Imputer": ["Imputer", "X", "y"],
+            "Undersampler": ["Undersampler", "Imputer.x", "y"],
+            "Logistic Regression Classifier": [
+                "Logistic Regression Classifier",
+                "Undersampler.x",
+                "Undersampler.y",
+            ],
+        }
+    )
+    component_graph1.instantiate({"Imputer": {"numeric_impute_strategy": "mean"}})
+    assert (
+        component_graph1.component_instances["Imputer"].parameters[
+            "numeric_impute_strategy"
+        ]
+        == "mean"
+    )
+
+    # make sure the value gets overwritten when reinitialized
+    bcp = BinaryClassificationPipeline(
+        component_graph1, parameters={"Imputer": {"numeric_impute_strategy": "median"}}
+    )
+    assert bcp.parameters["Imputer"]["numeric_impute_strategy"] == "median"
+    assert (
+        bcp.component_graph.component_instances["Imputer"].parameters[
+            "numeric_impute_strategy"
+        ]
+        == "median"
+    )