MAINT: Fix missed usage of deprecated tools.rank

statsmodels/regression/linear_model.py

@@ -44,8 +44,8 @@
 from scipy import optimize
 from scipy.stats import chi2
 
-from statsmodels.tools.tools import (add_constant, rank,
-                                     recipr, chain_dot, pinv_extended)
+from statsmodels.compatnp.np_compat import np_matrix_rank
+from statsmodels.tools.tools import add_constant, chain_dot, pinv_extended
 from statsmodels.tools.decorators import (resettable_cache,
                                           cache_readonly,
                                           cache_writable)
@@ -110,7 +110,7 @@ def df_model(self):
         """
         if self._df_model is None:
             if self.rank is None:
-                self.rank = rank(self.exog)
+                self.rank = np_matrix_rank(self.exog)
             self._df_model = float(self.rank - self.k_constant)
         return self._df_model
 
@@ -127,7 +127,7 @@ def df_resid(self):
 
         if self._df_resid is None:
             if self.rank is None:
-                self.rank = rank(self.exog)
+                self.rank = np_matrix_rank(self.exog)
             self._df_resid = self.nobs - self.rank
         return self._df_resid
 
@@ -151,7 +151,7 @@ def _has_constant(self):
             return True
         # Compute rank of augmented matrix
         augmented_exog = add_constant(self.exog)
-        return rank(augmented_exog)==rank(self.exog)
+        return np_matrix_rank(augmented_exog) == np_matrix_rank(self.exog)
 
     def whiten(self, X):
         raise NotImplementedError("Subclasses should implement.")
@@ -194,7 +194,7 @@ def fit(self, method="pinv", **kwargs):
 
                 # Cache these singular values for use later.
                 self.wexog_singular_values = singular_values
-                self.rank = rank(np.diag(singular_values))
+                self.rank = np_matrix_rank(np.diag(singular_values))
 
             beta = np.dot(self.pinv_wexog, self.wendog)
 
@@ -209,7 +209,7 @@ def fit(self, method="pinv", **kwargs):
 
                 # Cache singular values from R.
                 self.wexog_singular_values = np.linalg.svd(R, 0, 0)
-                self.rank = rank(R)
+                self.rank = np_matrix_rank(R)
             else:
                 Q, R = self.exog_Q, self.exog_R
 

statsmodels/regression/quantile_regression.py

@@ -22,7 +22,8 @@
 import scipy.stats as stats
 from scipy.linalg import pinv
 from scipy.stats import norm
-from statsmodels.tools.tools import chain_dot, rank
+from statsmodels.tools.tools import chain_dot
+from statsmodels.compatnp.np_compat import np_matrix_rank
 from statsmodels.tools.decorators import cache_readonly
 from statsmodels.regression.linear_model import (RegressionModel,
                                                  RegressionResults,
@@ -137,7 +138,7 @@ def fit(self, q=.5, vcov='robust', kernel='epa', bandwidth='hsheather',
         endog = self.endog
         exog = self.exog
         nobs = self.nobs
-        exog_rank = rank(self.exog)
+        exog_rank = np_matrix_rank(self.exog)
         self.rank = exog_rank
         self.df_model = float(self.rank - self.k_constant)
         self.df_resid = self.nobs - self.rank

statsmodels/regression/tests/test_regression.py

@@ -9,7 +9,8 @@
 from numpy.testing import (assert_almost_equal, assert_approx_equal,
                             assert_raises, assert_equal, assert_allclose)
 from scipy.linalg import toeplitz
-from statsmodels.tools.tools import add_constant, categorical, rank
+from statsmodels.tools.tools import add_constant, categorical
+from statsmodels.compatnp.np_compat import np_matrix_rank
 from statsmodels.regression.linear_model import OLS, WLS, GLS, yule_walker
 from statsmodels.datasets import longley
 from scipy.stats import t as student_t
@@ -162,7 +163,7 @@ def setupClass(cls):
         Q, R = np.linalg.qr(data.exog)
         model_qr.exog_Q, model_qr.exog_R  = Q, R
         model_qr.normalized_cov_params = np.linalg.inv(np.dot(R.T, R))
-        model_qr.rank = rank(R)
+        model_qr.rank = np_matrix_rank(R)
         res_qr2 = model_qr.fit(method="qr")
 
         cls.res_qr = res_qr

statsmodels/stats/contrast.py

@@ -2,7 +2,8 @@
 from scipy.stats import f as fdist
 from scipy.stats import t as student_t
 from scipy import stats
-from statsmodels.tools.tools import clean0, rank, fullrank
+from statsmodels.tools.tools import clean0, fullrank
+from statsmodels.compatnp.np_compat import np_matrix_rank
 
 
 #TODO: should this be public if it's just a container?
@@ -285,7 +286,7 @@ def contrastfromcols(L, D, pseudo=None):
     if len(Lp.shape) == 1:
         Lp.shape = (n, 1)
 
-    if rank(Lp) != Lp.shape[1]:
+    if np_matrix_rank(Lp) != Lp.shape[1]:
         Lp = fullrank(Lp)
         C = np.dot(pseudo, Lp).T
 

statsmodels/tools/catadd.py

@@ -1,7 +1,5 @@
-
-
 import numpy as np
-from statsmodels.tools.tools import rank as smrank
+from statsmodels.compatnp.np_compat import np_matrix_rank
 
 
 def add_indep(x, varnames, dtype=None):
@@ -31,7 +29,7 @@ def add_indep(x, varnames, dtype=None):
     for (xi, ni) in zip(x, varnames):
         #print xi.shape, xout.shape
         xout[:,count] = xi
-        rank_new = smrank(xout)
+        rank_new = np_matrix_rank(xout)
         #print rank_new
         if  rank_new > rank_old:
             varnames_new.append(ni)