Merge pull request #6478 from bashtage/0.11.cherry

MAINT: Cherry pick recent pulls for 0.11

lint.sh

@@ -90,6 +90,7 @@ if [ "$LINT" == true ]; then
         statsmodels/graphics/api.py \
         statsmodels/graphics/functional.py \
         statsmodels/graphics/tests/test_agreement.py \
+        statsmodels/graphics/tests/test_boxplots.py \
         statsmodels/graphics/tests/test_correlation.py \
         statsmodels/graphics/tests/test_functional.py \
         statsmodels/graphics/tests/test_gofplots.py \

pyproject.toml

@@ -5,7 +5,7 @@ requires = [
     "cython>=0.29.14",
     "numpy==1.14.5; python_version=='3.5'",
     "numpy==1.14.5; python_version=='3.6'",
-    "numpy==1.14.5; python_version>='3.7'",
+    "numpy==1.16.1; python_version=='3.7'",
     "numpy==1.17.5; python_version>='3.8'",
     "scipy>=1.0",
 ]

setup.cfg

@@ -42,6 +42,7 @@ filterwarnings =
     error:The default value of lags:FutureWarning
     error::pandas.core.common.SettingWithCopyWarning
     error:non-integer arg n is deprecated:DeprecationWarning
+    error:Creating an ndarray:numpy.VisibleDeprecationWarning
 markers =
     example: mark a test that runs example code
     matplotlib: mark a test that requires matplotlib

statsmodels/graphics/boxplots.py

@@ -13,8 +13,9 @@
 
 
 def violinplot(data, ax=None, labels=None, positions=None, side='both',
-               show_boxplot=True, plot_opts={}):
-    """Make a violin plot of each dataset in the `data` sequence.
+               show_boxplot=True, plot_opts=None):
+    """
+    Make a violin plot of each dataset in the `data` sequence.
 
     A violin plot is a boxplot combined with a kernel density estimate of the
     probability density function per point.
@@ -123,8 +124,8 @@ def violinplot(data, ax=None, labels=None, positions=None, side='both',
 
     .. plot:: plots/graphics_boxplot_violinplot.py
     """
-
-    if np.size(data) == 0:
+    plot_opts = {} if plot_opts is None else plot_opts
+    if max([np.size(arr) for arr in data]) == 0:
         msg = "No Data to make Violin: Try again!"
         raise ValueError(msg)
 
@@ -141,8 +142,7 @@ def violinplot(data, ax=None, labels=None, positions=None, side='both',
 
     # Plot violins.
     for pos_data, pos in zip(data, positions):
-        xvals, violin = _single_violin(ax, pos, pos_data, width, side,
-                                       plot_opts)
+        _single_violin(ax, pos, pos_data, width, side, plot_opts)
 
     if show_boxplot:
         ax.boxplot(data, notch=1, positions=positions, vert=1)

statsmodels/graphics/tests/test_boxplots.py

@@ -10,61 +10,89 @@
     pass
 
 
-@pytest.mark.matplotlib
-def test_violinplot_beanplot(close_figures):
+@pytest.fixture(scope="module")
+def age_and_labels():
     # Test violinplot and beanplot with the same dataset.
     data = anes96.load_pandas()
     party_ID = np.arange(7)
     labels = ["Strong Democrat", "Weak Democrat", "Independent-Democrat",
               "Independent-Independent", "Independent-Republican",
               "Weak Republican", "Strong Republican"]
-
     age = [data.exog['age'][data.endog == id] for id in party_ID]
+    return age, labels
 
-    fig = plt.figure()
-    ax = fig.add_subplot(111)
+
+@pytest.mark.matplotlib
+def test_violinplot(age_and_labels, close_figures):
+    age, labels = age_and_labels
+
+    fig, ax = plt.subplots(1, 1)
     violinplot(age, ax=ax, labels=labels,
-               plot_opts={'cutoff_val':5, 'cutoff_type':'abs',
-                          'label_fontsize':'small',
-                          'label_rotation':30})
+               plot_opts={'cutoff_val': 5, 'cutoff_type': 'abs',
+                          'label_fontsize': 'small',
+                          'label_rotation': 30})
+
 
-    fig = plt.figure()
-    ax = fig.add_subplot(111)
+@pytest.mark.matplotlib
+def test_violinplot_bw_factor(age_and_labels, close_figures):
+    age, labels = age_and_labels
+
+    fig, ax = plt.subplots(1, 1)
     violinplot(age, ax=ax, labels=labels,
-               plot_opts={'cutoff_val':5, 'cutoff_type':'abs',
-                          'label_fontsize':'small',
-                          'label_rotation':30,
-                          'bw_factor':.2})
+               plot_opts={'cutoff_val': 5, 'cutoff_type': 'abs',
+                          'label_fontsize': 'small',
+                          'label_rotation': 30,
+                          'bw_factor': .2})
+
+
+@pytest.mark.matplotlib
+def test_beanplot(age_and_labels, close_figures):
+    age, labels = age_and_labels
 
-    fig = plt.figure()
-    ax = fig.add_subplot(111)
+    fig, ax = plt.subplots(1, 1)
     beanplot(age, ax=ax, labels=labels,
-             plot_opts={'cutoff_val':5, 'cutoff_type':'abs',
-                        'label_fontsize':'small',
-                        'label_rotation':30})
+             plot_opts={'cutoff_val': 5, 'cutoff_type': 'abs',
+                        'label_fontsize': 'small',
+                        'label_rotation': 30})
+
+
+@pytest.mark.matplotlib
+def test_beanplot_jitter(age_and_labels, close_figures):
+    age, labels = age_and_labels
 
-    fig = plt.figure()
-    ax = fig.add_subplot(111)
+    fig, ax = plt.subplots(1, 1)
     beanplot(age, ax=ax, labels=labels, jitter=True,
              plot_opts={'cutoff_val': 5, 'cutoff_type': 'abs',
                         'label_fontsize': 'small',
                         'label_rotation': 30})
 
-    fig = plt.figure()
-    ax = fig.add_subplot(111)
+
+@pytest.mark.matplotlib
+def test_beanplot_side_right(age_and_labels, close_figures):
+    age, labels = age_and_labels
+
+    fig, ax = plt.subplots(1, 1)
     beanplot(age, ax=ax, labels=labels, jitter=True, side='right',
              plot_opts={'cutoff_val': 5, 'cutoff_type': 'abs',
                         'label_fontsize': 'small',
                         'label_rotation': 30})
 
-    fig = plt.figure()
-    ax = fig.add_subplot(111)
+
+@pytest.mark.matplotlib
+def test_beanplot_side_left(age_and_labels, close_figures):
+    age, labels = age_and_labels
+
+    fig, ax = plt.subplots(1, 1)
     beanplot(age, ax=ax, labels=labels, jitter=True, side='left',
              plot_opts={'cutoff_val': 5, 'cutoff_type': 'abs',
                         'label_fontsize': 'small',
                         'label_rotation': 30})
 
-    fig = plt.figure()
-    ax = fig.add_subplot(111)
+
+@pytest.mark.matplotlib
+def test_beanplot_legend_text(age_and_labels, close_figures):
+    age, labels = age_and_labels
+
+    fig, ax = plt.subplots(1, 1)
     beanplot(age, ax=ax, labels=labels,
              plot_opts={'bean_legend_text': 'text'})

statsmodels/nonparametric/bandwidths.py

@@ -170,7 +170,7 @@ def select_bandwidth(x, bw, kernel):
     bandwidth = bandwidth_funcs[bw](x, kernel)
     if bandwidth == 0:
         # eventually this can fall back on another selection criterion.
-        err = "Selected KDE bandwidth is 0. Cannot estiamte density."
+        err = "Selected KDE bandwidth is 0. Cannot estimate density."
         raise RuntimeError(err)
     else:
         return bandwidth

statsmodels/regression/linear_model.py

@@ -1776,6 +1776,9 @@ def fvalue(self):
     @cache_readonly
     def f_pvalue(self):
         """The p-value of the F-statistic."""
+        # Special case for df_model 0
+        if self.df_model == 0:
+            return np.full_like(self.fvalue, np.nan)
         return stats.f.sf(self.fvalue, self.df_model, self.df_resid)
 
     @cache_readonly

statsmodels/regression/tests/test_regression.py

@@ -1351,3 +1351,13 @@ def test_bool_regressor(reset_randomstate):
     bool_res = OLS(endog, exog).fit()
     res = OLS(endog, exog.astype(np.double)).fit()
     assert_allclose(bool_res.params, res.params)
+
+
+def test_ols_constant(reset_randomstate):
+    y = np.random.standard_normal((200))
+    x = np.ones((200, 1))
+    res = OLS(y, x).fit()
+    with pytest.warns(None) as recording:
+        assert np.isnan(res.fvalue)
+        assert np.isnan(res.f_pvalue)
+    assert len(recording) == 0

statsmodels/stats/libqsturng/qsturng_.py

@@ -821,20 +821,23 @@ def _psturng(q, r, v):
     if q < 0.:
         raise ValueError('q should be >= 0')
 
-    opt_func = lambda p, r, v : abs(_qsturng(p, r, v) - q)
+    def opt_func(p, r, v):
+        return np.squeeze(abs(_qsturng(p, r, v) - q))
 
     if v == 1:
         if q < _qsturng(.9, r, 1):
             return .1
         elif q > _qsturng(.999, r, 1):
             return .001
-        return 1. - fminbound(opt_func, .9, .999, args=(r,v))
+        soln = 1. - fminbound(opt_func, .9, .999, args=(r,v))
+        return np.atleast_1d(soln)
     else:
         if q < _qsturng(.1, r, v):
             return .9
         elif q > _qsturng(.999, r, v):
             return .001
-        return 1. - fminbound(opt_func, .1, .999, args=(r,v))
+        soln = 1. - fminbound(opt_func, .1, .999, args=(r,v))
+        return np.atleast_1d(soln)
 
 _vpsturng = np.vectorize(_psturng)
 _vpsturng.__doc__ = """vector version of psturng"""

statsmodels/stats/tests/test_pairwise.py

@@ -258,11 +258,14 @@ def setup_class(cls):
 
     def test_incorrect_output(self):
         # too few groups
-        assert_raises(ValueError, MultiComparison, np.array([1] * 10), [1, 2] * 4)
+        with pytest.raises(ValueError):
+            MultiComparison(np.array([1] * 10), [1, 2] * 4)
         # too many groups
-        assert_raises(ValueError, MultiComparison, np.array([1] * 10), [1, 2] * 6)
+        with pytest.raises(ValueError):
+            MultiComparison(np.array([1] * 10), [1, 2] * 6)
         # just one group
-        assert_raises(ValueError, MultiComparison, np.array([1] * 10), [1] * 10)
+        with pytest.raises(ValueError):
+            MultiComparison(np.array([1] * 10), [1] * 10)
 
         # group_order does not select all observations, only one group left
         with warnings.catch_warnings(record=True) as w:

statsmodels/tsa/innovations/tests/test_cython_arma_innovations_fast.py

@@ -38,10 +38,9 @@ def test_brockwell_davis_ex533():
     # through by sigma^2
     arma_process_acovf /= sigma2
     unconditional_variance /= sigma2
-    out = np.array(_arma_innovations.darma_transformed_acovf_fast(
-        ar, ma, arma_process_acovf))
-    acovf = np.array(out[0])
-    acovf2 = np.array(out[1])
+    transformed_acovf = _arma_innovations.darma_transformed_acovf_fast(
+        ar, ma, arma_process_acovf)
+    acovf, acovf2 = (np.array(arr) for arr in transformed_acovf)
 
     # `acovf` is an m^2 x m^2 matrix, where m = max(p, q)
     # but it is only valid for the autocovariances of the first m observations
@@ -144,10 +143,9 @@ def test_brockwell_davis_ex534():
                     [7.17133, 6.44139, 5.06027], atol=1e-5)
 
     # Next, get the autocovariance of the transformed process
-    out = np.array(_arma_innovations.darma_transformed_acovf_fast(
-        ar, ma, arma_process_acovf))
-    acovf = np.array(out[0])
-    acovf2 = np.array(out[1])
+    transformed_acovf = _arma_innovations.darma_transformed_acovf_fast(
+        ar, ma, arma_process_acovf)
+    acovf, acovf2 = (np.array(arr) for arr in transformed_acovf)
     # See test_brockwell_davis_ex533 for details on acovf vs acovf2
 
     # Test acovf
@@ -226,8 +224,9 @@ def test_innovations_algo_filter_kalman_filter(ar_params, ma_params, sigma2):
 
     # Innovations algorithm approach
     arma_process_acovf = arma_acovf(ar, ma, nobs=nobs, sigma2=sigma2)
-    acovf, acovf2 = np.array(_arma_innovations.darma_transformed_acovf_fast(
-                     ar, ma, arma_process_acovf / sigma2))
+    transformed_acov = _arma_innovations.darma_transformed_acovf_fast(
+        ar, ma, arma_process_acovf / sigma2)
+    acovf, acovf2 = (np.array(mv) for mv in transformed_acov)
     theta, r = _arma_innovations.darma_innovations_algo_fast(
         nobs, ar_params, ma_params, acovf, acovf2)
     u = _arma_innovations.darma_innovations_filter(endog, ar_params, ma_params,

statsmodels/tsa/tests/test_arima.py

@@ -2686,7 +2686,7 @@ def test_arma_repeated_fit():
     arma = ARMA(x, (1, 1))
     res = arma.fit(trend='c', disp=-1)
     repeat = arma.fit(trend='c', disp=-1)
-    rtol = 1e-5 if PLATFORM_WIN32 else 1e-7
+    rtol = 1e-4 if PLATFORM_WIN32 else 1e-7
     assert_allclose(res.params, repeat.params, rtol=rtol)
     assert isinstance(res.summary().as_text(), str)
     assert isinstance(repeat.summary().as_text(), str)

tools/ci/azure_template.yml

@@ -22,6 +22,8 @@ jobs:
         python.version: '3.6'
       Python37:
         python.version: '3.7'
+      Python38:
+        python.version: '3.8'
     maxParallel: 4
 
   steps:
@@ -37,7 +39,7 @@ jobs:
     displayName: 'Install dependencies'
 
   - script: |
-      python setup.py build_ext --inplace -q
+      pip install -e .
     displayName: 'Build Cython Extensions'
 
   - script: |

tools/ci/docbuild.sh

@@ -65,9 +65,11 @@ if [[ -z "$TRAVIS_TAG" ]]; then
   doctr deploy --built-docs docs/build/html/ --deploy-repo statsmodels/statsmodels.github.io devel > /dev/null;
 else
   if [[ "$TRAVIS_TAG" != *"dev"* ]]; then  # do not push on dev tags
-    doctr deploy --build-tags --built-docs docs/build/html/ --deploy-repo statsmodels/statsmodels.github.io "$TRAVIS_TAG";
+    echo doctr deploy --build-tags --built-docs docs/build/html/ --deploy-repo statsmodels/statsmodels.github.io "$TRAVIS_TAG" > /dev/null;
+    doctr deploy --build-tags --built-docs docs/build/html/ --deploy-repo statsmodels/statsmodels.github.io "$TRAVIS_TAG" > /dev/null;
     if [[ "$TRAVIS_TAG" != *"rc"* ]]; then  # do not push on main on rc
-      doctr deploy --build-tags --built-docs docs/build/html/ --deploy-repo statsmodels/statsmodels.github.io stable;
+      echo doctr deploy --build-tags --built-docs docs/build/html/ --deploy-repo statsmodels/statsmodels.github.io stable > /dev/null;
+      doctr deploy --build-tags --built-docs docs/build/html/ --deploy-repo statsmodels/statsmodels.github.io stable > /dev/null;
     fi;
   fi;
 fi;