From ccc8598613abf11c7e98738d756282f62905b5fe Mon Sep 17 00:00:00 2001
From: bastien-mva <bastien.batardiere@gmail.com>
Date: Wed, 26 Jun 2024 11:01:04 +0200
Subject: [PATCH 1/2] add summary on the latent variables.
---
pyPLNmodels/microcosm.py | 4 +--
pyPLNmodels/models.py | 54 ++++++++++++++++++++++++++++++++++++++++
tests/test_common.py | 9 +++----
3 files changed, 60 insertions(+), 7 deletions(-)
diff --git a/pyPLNmodels/microcosm.py b/pyPLNmodels/microcosm.py
index 7f2575cc..95d63cf4 100644
--- a/pyPLNmodels/microcosm.py
+++ b/pyPLNmodels/microcosm.py
@@ -26,9 +26,9 @@ def load_microcosm(
Parameters
----------
n_samples : int, optional
- Number of samples, by default max_samples.
+ Number of samples, by default 300.
dim : int, optional
- Dimension, by default max_dim.
+ Dimension, by default 200.
get_affil: bool, optional (keyword-only)
If True, will return the affiliations also. Default to False .
cov_list: list, optional (keyword-only).
diff --git a/pyPLNmodels/models.py b/pyPLNmodels/models.py
index a8deabcc..182c5992 100644
--- a/pyPLNmodels/models.py
+++ b/pyPLNmodels/models.py
@@ -13,6 +13,7 @@ import matplotlib.pyplot as plt
from sklearn.decomposition import PCA
import matplotlib
from scipy import stats
+from statsmodels.api import OLS
from pyPLNmodels._closed_forms import (
_closed_formula_coef,
@@ -459,6 +460,51 @@ class _model(ABC):
self._print_end_of_fitting_message(stop_condition, tol)
self._fitted = True
+ def summary(
+ self,
+ variable_number,
+ yname: str = None,
+ xname: list[str] = None,
+ title: str = None,
+ alpha: float = 0.05,
+ slim: bool = False,
+ ):
+ """
+ Summary from statsmodels on the latent variables.
+
+ parameters
+ ----------
+ yname : str, Optional
+ Name of endogenous (response) variable. The Default is y.
+ xname : str, Optional
+ Names for the exogenous variables. Default is var_## for ##
+ in the number of regressors.
+ Must match the number of parameters in the model.
+
+ title : str, Optional
+ Title for the top table. If not None, then this replaces the default title.
+ alpha : float, optional
+ The significance level for the confidence intervals.
+ slim: bool, Optional
+ Flag indicating to produce reduced set or diagnostic information. Default is False.
+ """
+ if self.exog is None:
+ print("No exog in the model, can not perform a summary.")
+ else:
+ ols = self._fit_ols(variable_number)
+ return ols.summary(
+ yname=yname, xname=xname, title=title, alpha=alpha, slim=slim
+ )
+
+ ## write docstrings on the summary function
+
+ def _fit_ols(self, variable_number):
+ return OLS(
+ self.latent_variables.numpy()[:, variable_number],
+ self.exog.numpy(),
+ hasconst=True,
+ ).fit()
+
@property
def fitted(self) -> bool:
"""
@@ -4577,6 +4623,14 @@ class ZIPln(_model):
variables = self.latent_prob
return self._viz_variables(variables, colors=colors, ax=ax, show_cov=False)
+ def _fit_ols(self, variable_number):
+ latent_variables, _ = self.latent_variables
+ return OLS(
+ latent_variables.numpy()[:, variable_number],
+ self.exog.numpy(),
+ hasconst=True,
+ ).fit()
+
class Brute_ZIPln(ZIPln):
@property
diff --git a/tests/test_common.py b/tests/test_common.py
index f4f05d61..450790df 100644
--- a/tests/test_common.py
+++ b/tests/test_common.py
@@ -12,10 +12,8 @@ pln_and_plnpca = ["Pln", "PlnPCA"]
single_models = ["Pln", "PlnPCA", "ZIPln"]
-# @pytest.mark.parametrize("any_model", dict_fixtures["loaded_and_fitted_model"])
-# @filter_models(single_models)
-@pytest.mark.parametrize("any_model", dict_fixtures["loaded_model"])
-@filter_models(["ZIPln"])
+@pytest.mark.parametrize("any_model", dict_fixtures["loaded_and_fitted_model"])
+@filter_models(single_models)
def test_properties(any_model):
assert hasattr(any_model, "latent_parameters")
assert hasattr(any_model, "latent_variables")
@@ -25,7 +23,7 @@ def test_properties(any_model):
@pytest.mark.parametrize("sim_model", dict_fixtures["loaded_and_fitted_model"])
@filter_models(pln_and_plnpca)
-def test_predict_simulated(sim_model):
+def test_predict_simulated_and_summary(sim_model):
if sim_model.nb_cov == 0:
assert sim_model.predict() is None
with pytest.raises(AttributeError):
@@ -35,6 +33,7 @@ def test_predict_simulated(sim_model):
prediction = sim_model.predict(X)
expected = X @ sim_model.coef
assert torch.all(torch.eq(expected, prediction))
+ print(sim_model.summary(variable_number=2))
@pytest.mark.parametrize("any_instance_model", dict_fixtures["instances"])
--
GitLab
From 2912a069a3cc52b052c85a62b0db8a5339221b5f Mon Sep 17 00:00:00 2001
From: bastien-mva <bastien.batardiere@gmail.com>
Date: Wed, 26 Jun 2024 11:07:19 +0200
Subject: [PATCH 2/2] add statsmodels.
---
pyproject.toml | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/pyproject.toml b/pyproject.toml
index ae5c5f4f..507915cb 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -53,6 +53,7 @@ dependencies = [
"scikit-learn",
"patsy",
"tqdm",
+"statsmodels",
]
[project.optional-dependencies]
tests = ["pytest","pytest_lazy_fixture"]
@@ -70,5 +71,3 @@ include-package-data = true
[tool.setuptools.packages.find]
include = ["pyPLNmodels", "pyPLNmodels.*"]
namespaces = false
-
-
--
GitLab