Create a minimal environment with core dependencies

.github/workflows/ci-tests.yaml

@@ -99,6 +99,7 @@ jobs:
       - name: Run tests against r
         run: |
           pixi run tests-against-r
+          pixi run tests-fixest-minimal
 
       - name: Upload coverage to Codecov (partial)
         uses: codecov/codecov-action@v4

README.md

@@ -56,8 +56,13 @@ You can install the release version from `PyPI` by running
 # inside an active virtual environment
 python -m pip install pyfixest
 ```
+To install a version with only core functionality and a reduced dependency set, you can run
 
-or the development version from github by running
+```py
+python -m pip install pyfixest[minimal]
+```
+
+You can install the development version from github by running
 
 ```py
 python -m pip install git+https://github.com/py-econometrics/pyfixest

pyfixest/estimation/decomposition.py

@@ -1,13 +1,34 @@
+import warnings
 from dataclasses import dataclass, field
 from typing import Any, Optional
 
 import numpy as np
 import pandas as pd
-from joblib import Parallel, delayed
 from numpy.typing import NDArray
 from scipy.sparse import hstack, spmatrix, vstack
 from scipy.sparse.linalg import lsqr
-from tqdm import tqdm
+
+try:
+    from tqdm import tqdm
+except ImportError:
+    warnings.warn(
+        "The tqdm package is not installed. Progress bars are disabled. Note that tqdm is included in the pyfixest default environment."
+    )
+
+    def tqdm(iterable, *args, **kwargs):
+        "Define a dummy tqdm function."
+        return iterable
+
+
+try:
+    from joblib import Parallel, delayed
+
+    joblib_available = True
+except ImportError:
+    joblib_available = False
+    warnings.warn(
+        "The joblib package is not installed. Parallel processing is disabled. Note that joblib is included in the pyfixest default environment."
+    )
 
 
 @dataclass
@@ -159,9 +180,12 @@ def bootstrap(self, rng: np.random.Generator, B: int = 1_000, alpha: float = 0.0
         if self.unique_clusters is not None:
             self.X_dict = {g: self.X_dict[g].tocsr() for g in self.X_dict}
 
-        _bootstrapped = Parallel(n_jobs=self.nthreads)(
-            delayed(self._bootstrap)(rng=rng) for _ in tqdm(range(B))
-        )
+        if joblib_available:
+            _bootstrapped = Parallel(n_jobs=self.nthreads)(
+                delayed(self._bootstrap)(rng=rng) for _ in tqdm(range(B))
+            )
+        else:
+            _bootstrapped = [self._bootstrap(rng=rng) for _ in tqdm(range(B))]
 
         self._bootstrapped = {
             key: np.concatenate([d[key] for d in _bootstrapped])

pyfixest/estimation/feols_compressed_.py

@@ -1,12 +1,24 @@
 import logging
+import warnings
 from collections.abc import Mapping
 from dataclasses import dataclass
 from typing import Any, Literal, Optional, Union
 
 import narwhals as nw
 import numpy as np
 import pandas as pd
-from tqdm import tqdm
+
+try:
+    from tqdm import tqdm
+except ImportError:
+    warnings.warn(
+        "The tqdm package is not installed. Progress bars are disabled. Note that tqdm is included in the pyfixest default environment."
+    )
+
+    def tqdm(iterable, *args, **kwargs):
+        "Define a dummy tqdm function."
+        return iterable
+
 
 from pyfixest.estimation.feols_ import Feols, PredictionErrorOptions, PredictionType
 from pyfixest.estimation.FormulaParser import FixestFormula

pyfixest/estimation/ritest.py

@@ -5,7 +5,7 @@
 import numba as nb
 import numpy as np
 import pandas as pd
-import seaborn as sns
+from scipy.stats import gaussian_kde
 
 # Make lets-plot an optional dependency
 try:
@@ -25,11 +25,24 @@
 except ImportError:
     _HAS_LETS_PLOT = False
 
+import warnings
+
 from scipy.stats import norm
-from tqdm import tqdm
 
 from pyfixest.estimation.demean_ import demean
 
+try:
+    from tqdm import tqdm
+except ImportError:
+    warnings.warn(
+        "The tqdm package is not installed. Progress bars are disabled. Note that tqdm is included in the pyfixest default environment."
+    )
+
+    def tqdm(iterable, *args, **kwargs):
+        "Define a dummy tqdm function."
+        return iterable
+
+
 # Only setup lets-plot if it's available
 if _HAS_LETS_PLOT:
     LetsPlot.setup_html()
@@ -382,10 +395,23 @@ def _plot_ritest_pvalue(
     x_lab = "Test statistic"
     y_lab = "Density"
 
+    def _plot_matplotlib(values: np.ndarray, sample_stat: np.ndarray) -> None:
+        kde = gaussian_kde(values)
+        x_range = np.linspace(min(values), max(values), 1000)
+        y_range = kde(x_range)
+
+        plt.figure(figsize=(10, 6))
+        plt.fill_between(x_range, y_range, color="blue", alpha=0.5)
+        plt.axvline(x=sample_stat, color="red", linestyle="--")
+        plt.title(title)
+        plt.xlabel(x_lab)
+        plt.ylabel(y_lab)
+        plt.show()
+
     if plot_backend == "lets_plot":
         if not _HAS_LETS_PLOT:
             print("lets-plot is not installed. Falling back to matplotlib.")
-            plot_backend = "matplotlib"
+            return _plot_matplotlib(df["ri_stats"].to_numpy(), sample_stat)
         else:
             plot = (
                 ggplot(df, aes(x="ri_stats"))
@@ -397,16 +423,10 @@ def _plot_ritest_pvalue(
                 + ylab(y_lab)
             )
 
-        return plot.show()
+            return plot.show()
 
     elif plot_backend == "matplotlib":
-        plt.figure(figsize=(10, 6))
-        sns.kdeplot(data=df, x="ri_stats", fill=True, color="blue", alpha=0.5)
-        plt.axvline(x=sample_stat, color="red", linestyle="--")
-        plt.title(title)
-        plt.xlabel(x_lab)
-        plt.ylabel(y_lab)
-        plt.show()
+        return _plot_matplotlib(df["ri_stats"].to_numpy(), sample_stat)
 
     else:
         raise ValueError(f"Unsupported plot backend: {plot_backend}")

pyfixest/report/summarize.py

@@ -6,8 +6,6 @@
 
 import numpy as np
 import pandas as pd
-from great_tables import GT
-from tabulate import tabulate
 
 from pyfixest.estimation.feiv_ import Feiv
 from pyfixest.estimation.feols_ import Feols
@@ -156,6 +154,8 @@ def etable(
     pf.etable([fit1, fit2])
     ```
     """
+    # check if etable installed else error
+
     if signif_code is None:
         signif_code = [0.001, 0.01, 0.05]
     assert isinstance(signif_code, list) and len(signif_code) == 3, (
@@ -704,6 +704,11 @@ def _tabulate_etable_md(df, n_coef, n_fixef, n_models, n_model_stats):
     -------
     - formatted_table (str): The formatted table as a string.
     """
+    try:
+        from tabulate import tabulate
+    except ImportError:
+        raise ImportError("The tabulate package is required but it is not installed.")
+
     # Format the DataFrame for tabulate
     table = tabulate(
         df,
@@ -1085,6 +1090,13 @@ def make_table(
             rowname_col = dfs.columns[0]
             groupname_col = None
 
+        try:
+            from great_tables import GT
+        except ImportError:
+            raise ImportError(
+                "The great_tables package is for running pf.etable() in 'gt' mode."
+            )
+
         # Generate the table with GT
         gt = GT(dfs, auto_align=False)
 

pyproject.toml

@@ -15,19 +15,27 @@ dependencies = [
   "formulaic>=1.1.0",
   "pandas>=1.1.0",
   "numba>=0.58.0",
-  "seaborn>=0.13.2",
-  "tabulate>=0.9.0",
-  "tqdm>=4.0.0",
-  "great-tables>=0.10.0",
   "numpy>=1.19.0",
   "narwhals>=1.13.3",
-  "joblib>=1.4.2,<2",
+  "matplotlib>=3.4",
 ]
 
 [tool.pixi.feature.dev.dependencies]
 rpy2 = ">=3.5.11,<4"
 
+[tool.pixi.feature.minimaldev.dependencies]
+rpy2 = ">=3.5.11,<4"
+
+
 [project.optional-dependencies]
+
+core = [
+  "joblib>=1.4.2,<2",
+  "tabulate>=0.9.0",
+  "tqdm>=4.0.0",
+  "great-tables>=0.10.0",
+]
+
 dev = [
   "pytest>=7.2.0",
   "pytest-cov>=4.1.0",
@@ -42,6 +50,13 @@ dev = [
   "pyarrow>=14.0",
   "jax>=0.4.15",
   "jaxlib>=0.4.15",
+  "seaborn>=0.13.2",
+]
+
+minimaldev = [
+  "pytest>=7.2.0",
+  "pytest-cov>=4.1.0",
+  "pytest-xdist>=3.5.0",
 ]
 
 plots = [
@@ -83,8 +98,10 @@ platforms = ["linux-64", "win-64", "osx-arm64", "osx-64"]
 pyfixest = { path = ".", editable = true }
 
 [tool.pixi.environments]
-default = { solve-group = "default" }
+minimal = { solve-group = "default" }
+default = {features = ["core"], solve-group = "default"}
 dev = { features = ["dev"], solve-group = "default" }
+minimaldev = { features = ["minimaldev"], solve-group = "default" }
 docs = { features = ["docs"], solve-group = "default" }
 build = { features = ["build"], solve-group = "default" }
 jax = { features = ["jax"], solve-group = "default" }
@@ -105,6 +122,10 @@ update-test-data = "Rscript tests/r_test_comparisons.R"
 install-r = "Rscript r_test_requirements.R"
 render-notebooks = "python scripts/run_notebooks.py"
 
+[tool.pixi.feature.minimaldev.tasks]
+tests-fixest-minimal = "pytest tests/test_vs_fixest.py::test_single_fit_feols -k 'not jax' --cov=pyfixest --cov-report=xml"
+
+
 [tool.pixi.feature.build.tasks]
 build-pip = 'python -m build .'
 
@@ -113,10 +134,6 @@ docs-build = "quartodoc build --verbose --config docs/_quarto.yml"
 docs-render = "quarto render docs"
 docs-preview = "quarto preview docs"
 
-[tool.pixi.feature.docs.dependencies]
-rpy2 = ">=3.5.11,<4"
-r-fixest = ">=0.12.1,<0.13"
-r-broom = ">=1.0.5,<2"
 
 [tool.pytest.ini_options]
 addopts = [

scripts/run_notebooks.py

@@ -4,11 +4,23 @@
 """
 
 import logging
+import warnings
 from pathlib import Path
 
 import papermill
 from joblib import Parallel, delayed
-from tqdm import tqdm
+
+try:
+    from tqdm import tqdm
+except ImportError:
+    warnings.warn(
+        "The tqdm package is not installed. Progress bars are disabled. Note that tqdm is included in the pyfixest default environment."
+    )
+
+    def tqdm(iterable, *args, **kwargs):
+        "Define a dummy tqdm function."
+        return iterable
+
 
 KERNEL_NAME: str = "python3"
 DOCS = Path("docs")