ENH Refactor Model Card

examples/plot_model_card.py

@@ -64,12 +64,6 @@
 # Then, we pass information other than metadata in kwargs.
 # We'll initialize a local repository and save the card with the model in it.
 
-limitations = "This model is not ready to be used in production."
-model_description = (
-    "This is a HistGradientBoostingClassifier model trained on breast cancer dataset."
-    " It's trained with Halving Grid Search Cross Validation, with parameter grids on"
-    " max_leaf_nodes and max_depth."
-)
 license = "mit"
 
 card_data = CardData(
@@ -79,21 +73,9 @@
     metrics=["acc"],
 )
 
-model_card_authors = "skops_user"
-get_started_code = (
-    "import pickle\nwith open(dtc_pkl_filename, 'rb') as file:\nclf = pickle.load(file)"
-)
-citation = "bibtex\n@inproceedings{...,year={2020}}"
-
-model_card = card.create_model_card(
-    model,
-    card_data=card_data,
-    limitations=limitations,
-    model_description=model_description,
-    citation_bibtex=citation,
-    model_card_authors=model_card_authors,
-    get_started_code=get_started_code,
-)
+
+model_card = card.Card(model)
+
 
 _, pkl_name = mkstemp(prefix="skops-", suffix=".pkl")
 
@@ -105,4 +87,20 @@
     model=pkl_name, requirements=[f"scikit-learn={sklearn.__version__}"], dst=local_repo
 )
 
-model_card.save(os.path.join(f"{local_repo}", "README.md"))
+limitations = "This model is not ready to be used in production."
+model_description = (
+    "This is a HistGradientBoostingClassifier model trained on breast cancer dataset."
+    " It's trained with Halving Grid Search Cross Validation, with parameter grids on"
+    " max_leaf_nodes and max_depth."
+)
+model_card_authors = "skops_user"
+get_started_code = (
+    "import pickle\nwith open(dtc_pkl_filename, 'rb') as file:\nclf = pickle.load(file)"
+)
+
+model_card.add("citation", "bibtex\n@inproceedings{...,year={2020}}")
+model_card.add("get_started_code", get_started_code).add(
+    "model_card_authors", model_card_authors
+)
+model_card.add("limitations", limitations).add("model_description", model_description)
+model_card.save(os.path.join(f"{local_repo}", "README.md"), card_data=card_data)

skops/card/_model_card.py

@@ -7,55 +7,65 @@
 import skops
 
 
-def _extract_estimator_config(model):
-    """Extracts estimator configuration and renders them into a vertical table.
-
-    Parameters
-    ----------
-        model (estimator): scikit-learn pipeline or model.
-
-    Returns
-    -------
-    str:
-        Markdown table of hyperparameters.
-    """
-    hyperparameter_dict = model.get_params(deep=True)
-    table = "| Hyperparameters | Value |\n| :-- | :-- |\n"
-    for hyperparameter, value in hyperparameter_dict.items():
-        table += f"| {hyperparameter} | {value} |\n"
-    return table
-
-
-def create_model_card(
-    model,
-    card_data,
-    **card_kwargs,
-):
-    """Creates a model card for the model and saves it to the target directory.
-
-    Parameters:
-    ----------
-    model: estimator
-        scikit-learn compatible estimator.
-    card_data: CardData
-        CardData object.
-    card_kwargs:
-        Card kwargs are information you can pass to fill in the sections of the
-        card template, e.g. model_description, citation_bibtex, get_started_code.
-    """
-    ROOT = skops.__path__
-    model_plot = re.sub(r"\n\s+", "", str(estimator_html_repr(model)))
-    hyperparameter_table = _extract_estimator_config(model)
-    card_data.library_name = "sklearn"
-    template_path = card_kwargs.get("template_path")
-    if template_path is None:
-        template_path = os.path.join(f"{ROOT[0]}", "card", "default_template.md")
-    card_kwargs["template_path"] = template_path
-    card = ModelCard.from_template(
-        card_data=card_data,
-        hyperparameter_table=hyperparameter_table,
-        model_plot=model_plot,
-        **card_kwargs,
-    )
-
-    return card
+class Card:
+    def __init__(self, model):
+        self.model = model
+        self.hyperparameters_table = self.extract_estimator_config(self.model)
+        self.model_plot = re.sub(r"\n\s+", "", str(estimator_html_repr(model)))
+        self.template_sections = {}
+
+    def add(self, section, value):
+        """Takes values to fill model card template.
+        Parameters:
+        ----------
+        section: str
+                Section in the template.
+        value: str
+                Value to fill in the section.
+        """
+        self.template_sections[section] = value
+        return self
+
+    def add_inspection(self):
+        pass
+
+    def save(self, path, card_data):
+        """Fills model card template, renders and saves it as markdown file to the target directory.
+
+        Parameters:
+        ----------
+        card_data: CardData
+            CardData object.
+        """
+        ROOT = skops.__path__
+        card_data.library_name = "sklearn"
+        template_path = self.template_sections.get("template_path")
+        if template_path is None:
+            template_path = os.path.join(f"{ROOT[0]}", "card", "default_template.md")
+        self.template_sections["template_path"] = template_path
+        card = ModelCard.from_template(
+            card_data=card_data,
+            hyperparameter_table=self.hyperparameter_table,
+            model_plot=self.model_plot,
+            **self.template_sections,
+        )
+        card.save(path)
+
+    def extract_estimator_config(model):
+        """Extracts estimator configuration and renders them into a vertical table.
+
+        Parameters
+        ----------
+            model (estimator): scikit-learn pipeline or model.
+
+        Returns
+        -------
+        str:
+            Markdown table of hyperparameters.
+        """
+
+        hyperparameter_dict = model.get_params(deep=True)
+        table = "| Hyperparameters | Value |\n| :-- | :-- |\n"
+        for hyperparameter, value in hyperparameter_dict.items():
+            table += f"| {hyperparameter} | {value} |\n"
+        return table