CLN: Remove pending deprecations

Remove non-tuple indexing
Remove convert_datetime64 flag
Fixed warning test to work with Python 3.7

.travis.yml

@@ -15,8 +15,12 @@ env:
     - USENUMPYBASE=false
     - MATPLOTLIB=
     - DOCBUILD=false
+    - MKL_NUM_THREADS=1 # Enforce single thread
+    - NUMEXPR_NUM_THREADS=1
+    - OMP_NUM_THREADS=1
+    - OPENBLAS_NUM_THREADS=1
     - PYTHONHASHSEED=0 # Ensure tests are correctly gathered by xdist
-    - OPEN_BLAS=false
+    - BLAS="mkl blas=*=mkl" # Use Intel MKL by default
 
 matrix:
   fast_finish: true
@@ -29,7 +33,7 @@ matrix:
   - python: 2.7
     env:
     - PYTHON=3.6
-    - OPEN_BLAS=true
+    - BLAS="nomkl blas=*=openblas"
     - COVERAGE=true
   # Python 2.7 + partially updated numpy, mpl; cutting edge scipy, pandas
   - python: 2.7
@@ -52,6 +56,7 @@ matrix:
     - PYTHON=2.7
     - NUMPY=1.11
     - USENUMPYBASE=true
+    - BLAS= # Do not specify blas in this config due to conflict
     - SCIPY=0.18
     - PANDAS=0.19
     - MATPLOTLIB=1.5
@@ -90,9 +95,6 @@ before_install:
   - chmod +x miniconda.sh
   - ./miniconda.sh -b -p /home/travis/miniconda
   - export PATH=/home/travis/miniconda/bin:$PATH
-  - export MKL_NUM_THREADS=1
-  - export NUMEXPR_NUM_THREADS=1
-  - export OMP_NUM_THREADS=1
   - conda config --set always_yes yes
   - conda update --quiet conda
   # Fix for headless TravisCI
@@ -102,18 +104,16 @@ before_install:
   - mkdir -p $HOME/.config/matplotlib
   - SRCDIR=$PWD
   # Build package list to avoid empty package=versions; only needed for versioned packages
-  - PKGS="python=${PYTHON}"
   - PKGS="numpy"; if [ ${NUMPY} ]; then PKGS="${PKGS}=${NUMPY}"; fi
   - if [ ${USENUMPYBASE} = true ]; then PKGS="${PKGS} numpy-base=${NUMPY}"; fi
   - PKGS="${PKGS} scipy"; if [ ${SCIPY} ]; then PKGS="${PKGS}=${SCIPY}"; fi
   - PKGS="${PKGS} patsy"; if [ ${PATSY} ]; then PKGS="${PKGS}=${PATSY}"; fi
   - PKGS="${PKGS} pandas"; if [ ${PANDAS} ]; then PKGS="${PKGS}=${PANDAS}"; fi
   - PKGS="${PKGS} Cython"; if [ ${CYTHON} ]; then PKGS="${PKGS}=${CYTHON}"; fi
   - if [ ${USEMPL} = true ]; then PKGS="${PKGS} matplotlib"; if [ ${MATPLOTLIB} ]; then PKGS="${PKGS}=${MATPLOTLIB}"; fi; fi
-  - if [ ${OPEN_BLAS} = true ]; then export export BLAS_CONFIG="nomkl blas=*=openblas"; else export BLAS_CONFIG="mkl blas=*=mkl"; fi
   - if [ ${COVERAGE} = true ]; then export COVERAGE_OPTS=" --cov-config=.travis_coveragerc --cov=statsmodels "; else export COVERAGE_OPTS=""; fi
-  - echo conda create --yes --quiet -n statsmodels-test ${PYTHON} ${BLAS_CONFIG} ${PKGS} ${OPTIONAL} pyyaml joblib
-  - conda create --yes --quiet -n statsmodels-test ${PYTHON} ${BLAS_CONFIG} ${PKGS} ${OPTIONAL} pyyaml joblib
+  - echo conda create --yes --quiet -n statsmodels-test python=${PYTHON} ${BLAS} ${PKGS} ${OPTIONAL} pyyaml joblib
+  - conda create --yes --quiet -n statsmodels-test python=${PYTHON} ${BLAS} ${PKGS} ${OPTIONAL} pyyaml joblib
   - source activate statsmodels-test
   - pip install 'pytest<4' pytest-xdist nose
   - if [ ${COVERAGE} = true ]; then pip install codecov coverage coveralls pytest-cov; fi

statsmodels/formula/tests/test_formula.py

@@ -165,7 +165,7 @@ def test_patsy_missing_data():
     with warnings.catch_warnings(record=True) as w:
         warnings.simplefilter("always")
         res2 = res.predict(data)
-        assert_equal(repr(w[-1].message),
-                     "ValueWarning('nan values have been dropped',)")
+        assert 'ValueWarning' in repr(w[-1].message)
+        assert 'nan values have been dropped' in repr(w[-1].message)
     # Frist record will be dropped in both cases
     assert_equal(res.fittedvalues, res2)

statsmodels/genmod/cov_struct.py

@@ -396,8 +396,8 @@ def initialize(self, model):
             # This is used to construct the working correlation
             # matrix.
             ilabel = np.zeros((ngrp, ngrp), dtype=np.int32)
-            ilabel[[ix1, ix2]] = ncm + 1
-            ilabel[[ix2, ix1]] = ncm + 1
+            ilabel[(ix1, ix2)] = ncm + 1
+            ilabel[(ix2, ix1)] = ncm + 1
             ilabels.append(ilabel)
 
             # This is used to estimate the variance components.

statsmodels/imputation/mice.py

@@ -1078,8 +1078,8 @@ def impute_pmm(self, vname):
 
         # Unwind the indices
         jj = np.arange(dxi.shape[0])
-        ix = dxi[[jj, ir]]
-        iz = ixm[[jj, ix]]
+        ix = dxi[(jj, ir)]
+        iz = ixm[(jj, ix)]
 
         imputed_miss = np.array(endog_obs[iz]).squeeze()
         self._store_changes(vname, imputed_miss)

statsmodels/iolib/summary2.py

@@ -223,7 +223,7 @@ def as_latex(self):
 
         if self._merge_latex:
             # create single tabular object for summary_col
-            tab = re.sub(to_replace,'\\midrule\n\\midrule\n', tab)
+            tab = re.sub(to_replace,'\\\midrule\n\\\midrule\n', tab)
 
         out = '\\begin{table}', title, tab, '\\end{table}'
         out = '\n'.join(out)

statsmodels/iolib/tests/test_table.py

@@ -198,8 +198,8 @@ def test_regression_with_tuples(self):
         df = df.join( i )
         endo = df.join( y )
         exo = df.join( x )
-        endo_groups = endo.groupby( ("i",) )
-        exo_groups = exo.groupby( ("i",) )
+        endo_groups = endo.groupby("i")
+        exo_groups = exo.groupby("i")
         exo_Df = exo_groups.agg( [np.sum, np.max] )
         endo_Df = endo_groups.agg( [np.sum, np.max] )
         reg = OLS(exo_Df[[("x", "sum")]],endo_Df).fit()

statsmodels/stats/_adnorm.py

@@ -61,8 +61,10 @@ def anderson_statistic(x, dist='norm', fit=True, params=(), axis=0):
     i = np.arange(1,N+1)
     sl1 = [None]*x.ndim
     sl1[axis] = slice(None)
+    sl1 = tuple(sl1)
     sl2 = [slice(None)]*x.ndim
     sl2[axis] = slice(None,None,-1)
+    sl2 = tuple(sl2)
     S = np.sum((2*i[sl1]-1.0)/N*(np.log(z)+np.log(1-z[sl2])), axis=axis)
     A2 = -N-S
     return A2

statsmodels/tools/eval_measures.py

@@ -303,7 +303,7 @@ def iqr(x1, x2, axis=0):
     idx = np.round((nobs-1) * np.array([0.25, 0.75])).astype(int)
     sl = [slice(None)] * xdiff.ndim
     sl[axis] = idx
-    iqr = np.diff(xdiff[sl], axis=axis)
+    iqr = np.diff(xdiff[tuple(sl)], axis=axis)
     iqr = np.squeeze(iqr)  # drop reduced dimension
     return iqr
 

statsmodels/tsa/statespace/representation.py

@@ -381,7 +381,7 @@ def __getitem__(self, key):
 
             # See note on time-varying arrays, below
             if matrix.shape[-1] == 1:
-                return matrix[[slice(None)]*(matrix.ndim-1) + [0]]
+                return matrix[tuple([slice(None)]*(matrix.ndim-1) + [0])]
             else:
                 return matrix
         # Otherwise if we have a tuple, we want a slice of a matrix

statsmodels/tsa/tests/test_stattools.py

@@ -148,7 +148,7 @@ def setup_class(cls):
         #cls.acf = np.concatenate(([1.], cls.acf))
         cls.qstat = cls.results['Q1']
         cls.res1 = acf(cls.x, nlags=40, qstat=True, alpha=.05)
-        cls.confint_res = cls.results[['acvar_lb','acvar_ub']].as_matrix()
+        cls.confint_res = cls.results[['acvar_lb','acvar_ub']].values
 
     def test_acf(self):
         assert_almost_equal(self.res1[0][1:41], self.acf, DECIMAL_8)

statsmodels/tsa/tests/test_tsa_tools.py

@@ -478,25 +478,25 @@ def test_dataframe(self):
         assert_frame_equal(expected, appended)
 
     def test_recarray(self):
-        recarray = pd.DataFrame(self.arr_2d).to_records(index=False, convert_datetime64=False)
+        recarray = pd.DataFrame(self.arr_2d).to_records(index=False)
         appended = tools.add_trend(recarray)
         expected = pd.DataFrame(self.arr_2d)
         expected['const'] = self.c
-        expected = expected.to_records(index=False, convert_datetime64=False)
+        expected = expected.to_records(index=False)
         assert_equal(expected, appended)
 
         prepended = tools.add_trend(recarray, prepend=True)
         expected = pd.DataFrame(self.arr_2d)
         expected.insert(0, 'const', self.c)
-        expected = expected.to_records(index=False, convert_datetime64=False)
+        expected = expected.to_records(index=False)
         assert_equal(expected, prepended)
 
         appended = tools.add_trend(recarray, trend='ctt')
         expected = pd.DataFrame(self.arr_2d)
         expected['const'] = self.c
         expected['trend'] = self.t
         expected['trend_squared'] = self.t ** 2
-        expected = expected.to_records(index=False, convert_datetime64=False)
+        expected = expected.to_records(index=False)
         assert_equal(expected, appended)
 
     def test_duplicate_const(self):

statsmodels/tsa/tsatools.py

@@ -115,7 +115,7 @@ def safe_is_const(s):
         x = np.column_stack(x[::order])
 
     if is_recarray:
-        x = x.to_records(index=False, convert_datetime64=False)
+        x = x.to_records(index=False)
         new_descr = x.dtype.descr
         extra_col = len(new_descr) - len(descr)
         if prepend: