BUG: Correct dimension when data removed

Correct exog dimension when data has been removed

closes statsmodels#6887

.travis.yml

@@ -57,7 +57,7 @@ matrix:
         - PYTHON=3.7
         - PANDAS=0.25
         - NUMPY=1.17
-        - SCIPY=1.3
+        - SCIPY=1.4
         - COVERAGE=true
         - PYTEST_OPTIONS=
     # Python 3.6 + legacy blas + older pandas
@@ -66,7 +66,7 @@ matrix:
         - PYTHON=3.6
         - NUMPY=1.16
         - PANDAS=0.24
-        - SCIPY=1.2
+        - SCIPY=1.3
         - BLAS="nomkl blas=*=openblas"
         - COVERAGE=true
         - PYTEST_OPTIONS=
@@ -75,7 +75,7 @@ matrix:
       env:
         - PYTHON=3.6
         - NUMPY=1.15
-        - SCIPY=1.1
+        - SCIPY=1.2
         - PANDAS=0.23
         - MATPLOTLIB=2
         - LINT=true

docs/source/install.rst

@@ -132,7 +132,7 @@ The current minimum dependencies are:
 
 * `Python <https://www.python.org>`__ >= 3.6
 * `NumPy <https://www.scipy.org/>`__ >= 1.15
-* `SciPy <https://www.scipy.org/>`__ >= 1.1
+* `SciPy <https://www.scipy.org/>`__ >= 1.2
 * `Pandas <https://pandas.pydata.org/>`__ >= 0.23
 * `Patsy <https://patsy.readthedocs.io/en/latest/>`__ >= 0.5.1
 

pyproject.toml

@@ -6,5 +6,5 @@ requires = [
     "numpy==1.15.4; python_version=='3.6'",
     "numpy==1.15.4; python_version=='3.7'",
     "numpy==1.17.5; python_version>='3.8'",
-    "scipy>=1.1",
+    "scipy>=1.2",
 ]

requirements.txt

@@ -1,4 +1,4 @@
 numpy>=1.15
-scipy>=1.1
+scipy>=1.2
 pandas>=0.23
 patsy>=0.5.1

statsmodels/regression/_prediction.py

@@ -157,7 +157,8 @@ def get_prediction(self, exog=None, transform=True, weights=None,
                 row_labels = None
 
         exog = np.asarray(exog)
-        if exog.ndim == 1 and (self.model.exog.ndim == 1 or
+        if exog.ndim == 1 and (self.model.exog is None or
+                               self.model.exog.ndim == 1 or
                                self.model.exog.shape[1] == 1):
             exog = exog[:, None]
         exog = np.atleast_2d(exog)  # needed in count model shape[1]

statsmodels/regression/tests/test_predict.py

@@ -258,3 +258,14 @@ def test_glm(self):
         assert_allclose(ci3b, ci3, rtol=1e-13)
         res_df = pred_res3b.summary_frame()
         assert_equal(res_df.index.values, [0, 1])
+
+
+def test_predict_remove_data():
+    # GH6887
+    endog = [i + np.random.normal(scale=0.1) for i in range(100)]
+    exog = [i for i in range(100)]
+    model = OLS(endog, exog, weights=[1 for _ in range(100)]).fit()
+    model.remove_data()
+    # works fine
+    assert_allclose(model.get_prediction(1).predicted_mean,
+                    model.get_prediction([1]).predicted_mean)

statsmodels/stats/contingency_tables.py

@@ -1333,7 +1333,7 @@ def mcnemar(table, exact=True, correction=True):
     if exact:
         statistic = np.minimum(n1, n2)
         # binom is symmetric with p=0.5
-        # SciPy 1.7+ required int arguments
+        # SciPy 1.7+ requires int arguments
         int_sum = int(n1 + n2)
         if int_sum != (n1 + n2):
             warnings.warn("exact can only be used with tables containing "

statsmodels/tests/test_package.py

@@ -1,10 +1,6 @@
 import subprocess
 import sys
 
-import pytest
-
-from statsmodels.compat.scipy import SCIPY_11
-
 
 def test_lazy_imports():
     # Check that when statsmodels.api is imported, matplotlib is _not_ imported
@@ -19,7 +15,6 @@ def test_lazy_imports():
     assert rc == 0
 
 
-@pytest.mark.skipif(SCIPY_11, reason='SciPy raises on -OO')
 def test_docstring_optimization_compat():
     # GH#5235 check that importing with stripped docstrings does not raise
     cmd = sys.executable + ' -OO -c "import statsmodels.api as sm"'

statsmodels/tsa/exponential_smoothing/initialization.py

@@ -46,7 +46,7 @@ def _initialization_heuristic(endog, trend=False, seasonal=False,
 
     # Seasonal component
     initial_seasonal = None
-    if seasonal is not None:
+    if seasonal:
         # Calculate the number of full cycles to use
         if nobs < 2 * seasonal_periods:
             raise ValueError('Cannot compute initial seasonals using'

statsmodels/tsa/statespace/tools.py

@@ -561,14 +561,15 @@ def _constrain_sv_less_than_one_python(unconstrained, order=None,
         Partial autocorrelation matrices. Should be a list of length
         `order`, where each element is an array sized `k_endog` x `k_endog`.
 
+    See Also
+    --------
+    constrain_stationary_multivariate
+
+
     Notes
     -----
     Corresponds to Lemma 2.2 in Ansley and Kohn (1986). See
     `constrain_stationary_multivariate` for more details.
-
-    There is a Cython implementation of this function that can be much faster,
-    but which requires SciPy 0.14.0 or greater. See
-    `constrain_stationary_multivariate` for details.
     """
 
     from scipy import linalg
@@ -620,14 +621,14 @@ def _compute_coefficients_from_multivariate_pacf_python(
         Transformed coefficient matrices leading to a stationary VAR
         representation.
 
+    See Also
+    --------
+    constrain_stationary_multivariate
+
     Notes
     -----
     Corresponds to Lemma 2.1 in Ansley and Kohn (1986). See
     `constrain_stationary_multivariate` for more details.
-
-    There is a Cython implementation of this function that can be much faster,
-    but which requires SciPy 0.14.0 or greater. See
-    `constrain_stationary_multivariate` for details.
     """
     from scipy import linalg