TST: Additional silence

Silence more tests

statsmodels/base/tests/test_shrink_pickle.py

@@ -258,7 +258,7 @@ def setup(self):
 class TestPickleFormula4(TestPickleFormula2):
 
     def setup(self):
-        self.results = sm.OLS.from_formula("Y ~ np.log(abs(A)+1) + B * C", data=self.data).fit()
+         self.results = sm.OLS.from_formula("Y ~ np.log(abs(A) + 1) + B * C", data=self.data).fit()
 
 # we need log in module namespace for the following test
 from numpy import log
@@ -267,7 +267,7 @@ class TestPickleFormula5(TestPickleFormula2):
     def setup(self):
         # if we import here, then unpickling fails -> exception in test
         #from numpy import log
-        self.results = sm.OLS.from_formula("Y ~ log(abs(A)+1) + B * C", data=self.data).fit()
+        self.results = sm.OLS.from_formula("Y ~ log(abs(A) + 1) + B * C", data=self.data).fit()
 
 
 class TestRemoveDataPicklePoissonRegularized(RemoveDataPickle):

statsmodels/discrete/tests/test_discrete.py

@@ -1585,7 +1585,8 @@ def test_formula_missing_exposure():
     assert_(type(mod1.exposure) is np.ndarray, msg='Exposure is not ndarray')
 
     # make sure this raises
-    exposure = pd.Series(np.random.chisquare(5, 5))
+    exposure = pd.Series(np.random.uniform(size=5))
+    df.loc[3, 'Bar'] = 4   # nan not relevant for ValueError for shape mismatch
     assert_raises(ValueError, sm.Poisson, df.Foo, df[['constant', 'Bar']],
                   exposure=exposure)
 

statsmodels/duration/survfunc.py

@@ -71,9 +71,9 @@ def _calc_survfunc_right(time, status, weights=None, entry=None, compress=True,
         se[(n == d) | (n == 0)] = np.nan
         se = np.cumsum(se)
         se = np.sqrt(se)
-        locs = np.logical_not(np.isinf(se) & (sp == 0))
+        locs = np.isfinite(se) | (sp != 0)
         se[locs] *= sp[locs]
-        se[np.logical_not(locs)] = np.nan
+        se[~locs] = np.nan
     else:
         # Tsiatis' (1981) formula
         se = d / (n * n).astype(np.float64)

statsmodels/nonparametric/tests/test_kernels.py

@@ -20,9 +20,9 @@
 
 y = results['accident']
 x = results['service']
-use_mask = x >= 0
-x = np.log(x[use_mask].copy())
-y = y[use_mask]
+positive = x >= 0
+x = np.log(x[positive])
+y = y[positive]
 xg = np.linspace(x.min(), x.max(), 40) # grid points default in Stata
 
 #kern_name = 'gau'

statsmodels/stats/contrast.py

@@ -55,12 +55,9 @@ def __init__(self, t=None, F=None, sd=None, effect=None, df_denom=None,
                 self.pvalue = self.dist.sf(self.statistic, df_denom)
             else:
                 "normal"
-                # TODO: Need a simpler method to handle NaNs
-                pvalue = np.empty_like(value)
-                pvalue.fill(np.nan)
-                not_nan = np.logical_not(np.isnan(value))
-                pvalue[not_nan] = self.dist.sf(np.abs(value[not_nan])) * 2
-                self.pvalue = pvalue
+                self.pvalue = np.full_like(value, np.nan)
+                not_nan = ~np.isnan(value)
+                self.pvalue[not_nan] = self.dist.sf(np.abs(value[not_nan])) * 2
 
         # cleanup
         # should we return python scalar?

statsmodels/stats/power.py

@@ -269,12 +269,18 @@ def func(x):
             success = 1
         else:
             # try backup
-            #TODO: check more cases to make this robust
-            val, infodict, ier, msg = optimize.fsolve(func, start_value,
-                                                      full_output=True) #scalar
-            #val = optimize.newton(func, start_value) #scalar
-            fval = infodict['fvec']
-            fit_res.append(infodict)
+            # TODO: check more cases to make this robust
+            if not np.isnan(start_value):
+                val, infodict, ier, msg = optimize.fsolve(func, start_value,
+                                                          full_output=True) #scalar
+                #val = optimize.newton(func, start_value) #scalar
+                fval = infodict['fvec']
+                fit_res.append(infodict)
+            else:
+                ier = -1
+                fval = 1
+                fit_res.append([None])
+
             if ier == 1 and np.abs(fval) < 1e-4 :
                 success = 1
             else:

statsmodels/stats/tests/test_tost.py

@@ -478,7 +478,7 @@ def test_tost_asym():
 
     #SMOKE tests: foe multi-endpoint vectorized, k on k
     resall = smws.ttost_ind(clinic[15:, 2:7], clinic[:15, 2:7],
-                           [-1.0, -1.0, -1.5, -1.5, -1.5], 0.6,
+                           np.exp([-1.0, -1.0, -1.5, -1.5, -1.5]), 0.6,
                            usevar='unequal', transform=np.log)
     resall = smws.ttost_ind(clinic[15:, 2:7], clinic[:15, 2:7],
                            [-1.0, -1.0, -1.5, -1.5, -1.5], 0.6,

statsmodels/tools/tools.py

@@ -364,7 +364,7 @@ def recipr(x):
     x = np.asarray(x)
     out = np.zeros_like(x, dtype=np.float64)
     nans = np.isnan(x.flat)
-    pos = np.logical_not(nans)
+    pos = ~nans
     pos[pos] = pos[pos] & (x.flat[pos] > 0)
     out.flat[pos] = 1.0 / x.flat[pos]
     out.flat[nans] = np.nan
@@ -380,9 +380,9 @@ def recipr0(x):
     x = np.asarray(x)
     out = np.zeros_like(x, dtype=np.float64)
     nans = np.isnan(x.flat)
-    not_zero = np.logical_not(nans)
-    not_zero[not_zero] = not_zero[not_zero] & (x.flat[not_zero] != 0)
-    out.flat[not_zero] = 1.0 / x.flat[not_zero]
+    non_zero = ~nans
+    non_zero[non_zero] = non_zero[non_zero] & (x.flat[non_zero] != 0)
+    out.flat[non_zero] = 1.0 / x.flat[non_zero]
     out.flat[nans] = np.nan
     return out