minor

Makefile

@@ -20,7 +20,7 @@ lint: ## run the lint checkers
 	uv run --all-extras ruff format tests
 
 install: ## install in poetry venv
-	uv pip install .
+	uv pip install -e .
 
 docs: readme ## build docs
-	uv run mkdocs build
+	uv run mkdocs build

marginaleffects/formulaic.py → marginaleffects/formula.py

@@ -6,7 +6,7 @@
 
 
 @validate_types
-def variables(formula: str):
+def get_variables(formula: str):
     """Extract variable names from a model formula.
 
     Parameters
@@ -28,9 +28,9 @@ def variables(formula: str):
 
     Examples
     --------
-    >>> variables("y ~ x1 + x2")
+    >>> get_variables("y ~ x1 + x2")
     ['y', 'x1', 'x2']
-    >>> variables("response ~ pred1 * pred2")
+    >>> get_variables("response ~ pred1 * pred2")
     ['response', 'pred1', 'pred2']
 
     Notes
@@ -77,14 +77,14 @@ def lwd(formula: str | None = None, vars: list[str] | None = None, data: pl.Data
     Either formula or vars must be provided to specify which variables to check.
     """
     if formula is not None:
-        vars = variables(formula)
+        vars = get_variables(formula)
     elif formula is None and vars is None:
         raise ValueError("formula or vars must be provided")
     return data.drop_nulls(subset=vars)
 
 
 @validate_types
-def design(formula: str, data: pl.DataFrame):
+def model_matrices(formula: str, data: pl.DataFrame):
     """Create design matrices from formula and data.
 
     Parameters
@@ -110,11 +110,11 @@ def design(formula: str, data: pl.DataFrame):
     Converts polars DataFrame to pandas for compatibility with formulaic.
     Flattens y to 1D array if response variable is categorical.
     """
-    vars = variables(formula)
+    vars = get_variables(formula)
     data = data.drop_nulls(subset=vars)
     y, X = formulaic.model_matrix(formula, data.to_pandas())
     y = np.ravel(data[vars[0]])  # avoid matrix if LHS is a character
-    return y, X, data
+    return y, X
 
 
 def _sanity_engine(engine):

marginaleffects/model_sklearn.py

@@ -1,7 +1,7 @@
 import numpy as np
 import warnings
 import polars as pl
-from . import formulaic as fml
+from .formula import get_variables
 from .model_abstract import ModelAbstract
 
 
@@ -99,7 +99,7 @@ def get_variables_names(self, variables=None, newdata=None):
         list
             Names of predictor variables, excluding the response variable.
         """
-        out = fml.variables(self.formula)[1:]
+        out = get_variables(self.formula)[1:]
         return out
 
     def get_predict(self, params, newdata: pl.DataFrame):

pyproject.toml

@@ -5,6 +5,7 @@ description = "Predictions, counterfactual comparisons, slopes, and hypothesis t
 readme = "README.md"
 requires-python = ">=3.10"
 dependencies = [
+    "formulaic >=1.0.2",
     "narwhals >=1.10.0",
     "numpy >=2.0.0",
     "patsy >=0.5.6",
@@ -16,7 +17,6 @@ dependencies = [
 [project.optional-dependencies]
 test = [
     "duckdb >=1.1.2",
-    "formulaic >=1.0.2",
     "matplotlib >=3.7.1",
     "pandas >=2.2.2",
     "pyarrow >=17.0.0",