Merge commit 'e417dc0fe26d5e87509678b680f5671eef613e1a' into backport…

…-0.14.1

setup.cfg

@@ -91,6 +91,8 @@ filterwarnings =
     error:The previous implementation of stack is deprecated:FutureWarning
     error:Series.__setitem__ treating keys as positions is deprecated:FutureWarning
     error:The provided callable:FutureWarning
+    error:divide by zero encountered in log1p:RuntimeWarning
+
 markers =
     example: mark a test that runs example code
     matplotlib: mark a test that requires matplotlib

statsmodels/stats/_adnorm.py

@@ -5,6 +5,8 @@
 Author: Josef Perktold and Scipy developers
 License : BSD-3
 """
+import warnings
+
 import numpy as np
 from scipy import stats
 
@@ -64,8 +66,12 @@ def anderson_statistic(x, dist='norm', fit=True, params=(), axis=0):
     sl2 = [slice(None)] * x.ndim
     sl2[axis] = slice(None, None, -1)
     sl2 = tuple(sl2)
-    s = np.sum((2 * i[sl1] - 1.0) / nobs * (np.log(z) + np.log1p(-z[sl2])),
-               axis=axis)
+    with warnings.catch_warnings():
+        warnings.filterwarnings(
+            "ignore", message="divide by zero encountered in log1p"
+        )
+        ad_values = (2 * i[sl1] - 1.0) / nobs * (np.log(z) + np.log1p(-z[sl2]))
+        s = np.sum(ad_values, axis=axis)
     a2 = -nobs - s
     return a2
 

statsmodels/stats/diagnostic.py

@@ -1072,7 +1072,7 @@ def linear_reset(res, power=3, test_type="fitted", use_f=False,
             raise ValueError("power must contains distinct integers all >= 2")
     exog = res.model.exog
     if test_type == "fitted":
-        aug = res.fittedvalues[:, None]
+        aug = np.asarray(res.fittedvalues)[:, None]
     elif test_type == "exog":
         # Remove constant and binary
         aug = res.model.exog
@@ -1293,7 +1293,7 @@ def linear_lm(resid, exog, func=None):
     if func is None:
         def func(x):
             return np.power(x, 2)
-
+    exog = np.asarray(exog)
     exog_aux = np.column_stack((exog, func(exog[:, 1:])))
 
     nobs, k_vars = exog.shape

statsmodels/stats/tests/test_diagnostic.py

@@ -1957,3 +1957,43 @@ def test_small_skip(reset_randomstate):
 # M2 + fit(M1)-exp(fit(M2)) < 2.22e-16 ***
 # ---
 # Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
+
+@pytest.mark.smoke
+def test_diagnostics_pandas(reset_randomstate):
+    # GH 8879
+    n = 100
+    df = pd.DataFrame(
+        {
+            "y": np.random.rand(n),
+            "x": np.random.rand(n),
+            "z": np.random.rand(n)}
+    )
+    y, x = df["y"], add_constant(df["x"])
+
+    res = OLS(df["y"], add_constant(df[["x"]])).fit()
+    res_large = OLS(df["y"], add_constant(df[["x", "z"]])).fit()
+    res_other = OLS(df["y"], add_constant(df[["z"]])).fit()
+    smsdia.linear_reset(res_large)
+    smsdia.linear_reset(res_large, test_type="fitted")
+    smsdia.linear_reset(res_large, test_type="exog")
+    smsdia.linear_reset(res_large, test_type="princomp")
+    smsdia.het_goldfeldquandt(y, x)
+    smsdia.het_breuschpagan(res.resid, x)
+    smsdia.het_white(res.resid, x)
+    smsdia.het_arch(res.resid)
+    smsdia.acorr_breusch_godfrey(res)
+    smsdia.acorr_ljungbox(y)
+    smsdia.linear_rainbow(res)
+    smsdia.linear_lm(res.resid, x)
+    smsdia.linear_harvey_collier(res)
+    smsdia.acorr_lm(res.resid)
+    smsdia.breaks_cusumolsresid(res.resid)
+    smsdia.breaks_hansen(res)
+    smsdia.compare_cox(res, res_other)
+    smsdia.compare_encompassing(res, res_other)
+    smsdia.compare_j(res, res_other)
+    smsdia.recursive_olsresiduals(res)
+    smsdia.recursive_olsresiduals(
+        res, order_by=np.arange(y.shape[0] - 1, 0 - 1, -1)
+    )
+    smsdia.spec_white(res.resid, x)