From b551e45e2bd1f9d8a955da948c16f14928f6eae0 Mon Sep 17 00:00:00 2001
From: Skipper Seabold <jsseabold@gmail.com>
Date: Wed, 24 Sep 2014 09:31:14 -0400
Subject: [PATCH 01/15] FIXUP/WIP: Refactor to use generic data handling

---
 statsmodels/regression/mixed_linear_model.py | 35 +++++++++++++++-----
 1 file changed, 27 insertions(+), 8 deletions(-)

diff --git a/statsmodels/regression/mixed_linear_model.py b/statsmodels/regression/mixed_linear_model.py
index 9cd8d7ace97..4e9f1c7c926 100644
--- a/statsmodels/regression/mixed_linear_model.py
+++ b/statsmodels/regression/mixed_linear_model.py
@@ -462,7 +462,7 @@ def __init__(self, endog, exog, groups, exog_re=None,
         # Calling super creates self.endog, etc. as ndarrays and the
         # original exog, endog, etc. are self.data.endog, etc.
         super(MixedLM, self).__init__(endog, exog, groups=groups,
-                                  exog_re=exog_re, missing=missing)
+                                      exog_re=exog_re, missing=missing)
 
         if exog_re is None:
             # Default random effects structure (random intercepts).
@@ -491,6 +491,8 @@ def __init__(self, endog, exog, groups, exog_re=None,
         # Override the default value
         self.nparams = self.k_fe + self.k_re2
 
+        self.data.xnames = self._make_param_names()
+
         # Convert the data to the internal representation, which is a
         # list of arrays, corresponding to the groups.
         group_labels = list(set(groups))
@@ -519,11 +521,23 @@ def __init__(self, endog, exog, groups, exog_re=None,
                                range(self.exog.shape[1])]
 
         # Set the random effect parameter names
-        if isinstance(self.exog_re, pd.DataFrame):
-            self.exog_re_names = list(self.exog_re.columns)
-        else:
-            self.exog_re_names = ["Z%d" % (k+1) for k in
-                                  range(self.exog_re.shape[1])]
+        #if isinstance(self.exog_re, pd.DataFrame):
+        #    self.exog_re_names = list(self.exog_re.columns)
+        #else:
+        #    self.exog_re_names = ["Z%d" % (k+1) for k in
+        #                          range(self.exog_re.shape[1])]
+
+    def _make_param_names(self):
+        names = list(self.exog_names)
+        jj = self.k_fe
+        for i in range(self.k_re):
+            for j in range(i + 1):
+                if i == j:
+                    names.append(self.exog_re_names[i] + " RE")
+                else:
+                    names.append(self.exog_re_names[j] + " x " +
+                                 self.exog_re_names[i] + " RE")
+                jj += 1
 
     @classmethod
     def from_formula(cls, formula, data, re_formula=None, subset=None,
@@ -756,7 +770,7 @@ def fit_regularized(self, start_params=None, method='l1', alpha=0,
         results.k_re = self.k_re
         results.k_re2 = self.k_re2
 
-        return results
+        return MixedLMResultsWrapper(results)
 
 
     def _reparam(self):
@@ -1685,7 +1699,7 @@ def fit(self, start_params=None, reml=True, niter_em=0,
         results.use_sqrt = self.use_sqrt
         results.freepat = self._freepat
 
-        return results
+        return MixedLMResultsWrapper(results)
 
 
 class MixedLMResults(base.LikelihoodModelResults):
@@ -1905,6 +1919,7 @@ def summary(self, yname=None, xname_fe=None, xname_re=None,
         jj = self.k_fe
         for i in range(self.k_re):
             for j in range(i + 1):
+                import ipdb; ipdb.set_trace()
                 if i == j:
                     names.append(self.model.exog_re_names[i] + " RE")
                 else:
@@ -2013,3 +2028,7 @@ def profile_re(self, re_ix, num_low=5, dist_low=1., num_high=5,
         model.exog_re_li = exog_re_li_save
 
         return likev
+
+
+class MixedLMResultsWrapper(base.LikelihoodResultsWrapper):
+    pass

From e109a32c5f7d2bbe7f3d20356723baccd614f251 Mon Sep 17 00:00:00 2001
From: Skipper Seabold <jsseabold@gmail.com>
Date: Thu, 25 Sep 2014 12:45:56 -0400
Subject: [PATCH 02/15] ENH: Allow more general param_names in data handling

---
 statsmodels/base/data.py | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)

diff --git a/statsmodels/base/data.py b/statsmodels/base/data.py
index 729fe49bd8e..6efe6bebdab 100644
--- a/statsmodels/base/data.py
+++ b/statsmodels/base/data.py
@@ -52,6 +52,8 @@ class ModelData(object):
     Class responsible for handling input data and extracting metadata into the
     appropriate form
     """
+    _param_names = None
+
     def __init__(self, endog, exog=None, missing='none', hasconst=None,
                  **kwargs):
         if missing != 'none':
@@ -251,6 +253,15 @@ def xnames(self):
             return list(xnames)
         return None
 
+    @property
+    def param_names(self):
+        # for handling names of 'extra' parameters in summary, etc.
+        return self._param_names or self.xnames
+
+    @param_names.setter
+    def param_names(self, values):
+        self._param_names = values
+
     @cache_readonly
     def row_labels(self):
         exog = self.orig_exog
@@ -383,9 +394,9 @@ def attach_columns(self, result):
         # don't squeeze because it might be a 2d row array
         # if it needs a squeeze, the bug is elsewhere
         if result.ndim <= 1:
-            return Series(result, index=self.xnames)
+            return Series(result, index=self.param_names)
         else:  # for e.g., confidence intervals
-            return DataFrame(result, index=self.xnames)
+            return DataFrame(result, index=self.param_names)
 
     def attach_columns_eq(self, result):
         return DataFrame(result, index=self.xnames, columns=self.ynames)

From da1ea47d8a0e59d847ed6fba1d95c78cc7394e7e Mon Sep 17 00:00:00 2001
From: Skipper Seabold <jsseabold@gmail.com>
Date: Thu, 25 Sep 2014 13:35:30 -0400
Subject: [PATCH 03/15] ENH: Allow attachment of generic names.

---
 statsmodels/base/data.py    | 12 +++++++++++-
 statsmodels/base/wrapper.py |  4 +++-
 2 files changed, 14 insertions(+), 2 deletions(-)

diff --git a/statsmodels/base/data.py b/statsmodels/base/data.py
index 6efe6bebdab..5c43347f3f5 100644
--- a/statsmodels/base/data.py
+++ b/statsmodels/base/data.py
@@ -313,7 +313,7 @@ def _check_integrity(self):
             if len(self.exog) != len(self.endog):
                 raise ValueError("endog and exog matrices are different sizes")
 
-    def wrap_output(self, obj, how='columns'):
+    def wrap_output(self, obj, how='columns', names=None):
         if how == 'columns':
             return self.attach_columns(obj)
         elif how == 'rows':
@@ -326,6 +326,8 @@ def wrap_output(self, obj, how='columns'):
             return self.attach_columns_eq(obj)
         elif how == 'cov_eq':
             return self.attach_cov_eq(obj)
+        elif how == 'generic_columns':
+            return self.attach_generic_columns(obj, names)
         else:
             return obj
 
@@ -347,6 +349,9 @@ def attach_rows(self, result):
     def attach_dates(self, result):
         return result
 
+    def attach_generic_columns(self, result, names):
+        return result
+
 
 class PatsyData(ModelData):
     def _get_names(self, arr):
@@ -389,6 +394,11 @@ def _get_row_labels(self, arr):
             # exog is not, so just return the row labels from endog
             return self.orig_endog.index
 
+    def attach_generic_columns(self, result, names):
+        # get the attribute to use
+        column_names = getattr(self, names, None)
+        return Series(result, index=column_names)
+
     def attach_columns(self, result):
         # this can either be a 1d array or a scalar
         # don't squeeze because it might be a 2d row array
diff --git a/statsmodels/base/wrapper.py b/statsmodels/base/wrapper.py
index 90572d82a33..c10246df170 100644
--- a/statsmodels/base/wrapper.py
+++ b/statsmodels/base/wrapper.py
@@ -35,7 +35,9 @@ def __getattribute__(self, attr):
         obj = getattr(results, attr)
         data = results.model.data
         how = self._wrap_attrs.get(attr)
-        if how:
+        if how and isinstance(how, tuple):
+            obj = data.wrap_output(obj, how=how[0], names=how[1])
+        elif how:
             obj = data.wrap_output(obj, how=how)
 
         return obj

From 812d6c5420f063c11e9221935b987936aeb18359 Mon Sep 17 00:00:00 2001
From: Skipper Seabold <jsseabold@gmail.com>
Date: Thu, 25 Sep 2014 13:41:15 -0400
Subject: [PATCH 04/15] ENH: Refactor to allow wrappers to work.

---
 statsmodels/regression/mixed_linear_model.py | 79 +++++++++++---------
 1 file changed, 42 insertions(+), 37 deletions(-)

diff --git a/statsmodels/regression/mixed_linear_model.py b/statsmodels/regression/mixed_linear_model.py
index 4e9f1c7c926..b91fc26393e 100644
--- a/statsmodels/regression/mixed_linear_model.py
+++ b/statsmodels/regression/mixed_linear_model.py
@@ -99,6 +99,7 @@
 import numpy as np
 import statsmodels.base.model as base
 from scipy.optimize import fmin_ncg, fmin_cg, fmin_bfgs, fmin
+from statsmodels.tools.decorators import cache_readonly
 from scipy.stats.distributions import norm
 import pandas as pd
 import patsy
@@ -108,6 +109,13 @@
 from statsmodels.tools.sm_exceptions import ConvergenceWarning
 from statsmodels.base._penalties import Penalty
 
+
+def _get_exog_re_names(exog_re):
+    if isinstance(exog_re, (pd.Series, pd.DataFrame)):
+        return exog_re.columns.tolist()
+    return ["Z{0}".format(k + 1) for k in range(exog_re.shape[1])]
+
+
 class MixedLMParams(object):
     """
     This class represents a parameter state for a mixed linear model.
@@ -454,6 +462,7 @@ def __init__(self, endog, exog, groups, exog_re=None,
         # If there is one covariate, it may be passed in as a column
         # vector, convert these to 2d arrays.
         # TODO: Can this be moved up in the class hierarchy?
+        #       yes, it should be done up the hierarchy
         if exog is not None and exog.ndim == 1:
             exog = exog[:,None]
         if exog_re is not None and exog_re.ndim == 1:
@@ -464,35 +473,35 @@ def __init__(self, endog, exog, groups, exog_re=None,
         super(MixedLM, self).__init__(endog, exog, groups=groups,
                                       exog_re=exog_re, missing=missing)
 
+        self.k_fe = exog.shape[1] # Number of fixed effects parameters
+
         if exog_re is None:
             # Default random effects structure (random intercepts).
+            self.k_re = 1
+            self.k_re2 = 1
             self.exog_re = np.ones((len(endog), 1), dtype=np.float64)
             self.data.exog_re = self.exog_re
+            self.data.param_names = self.exog_names + ['Intercept']
         else:
             # Process exog_re the same way that exog is handled
             # upstream
+            # TODO: this is wrong and should be handled upstream wholly
             self.data.exog_re = exog_re
             self.exog_re = np.asarray(exog_re)
-
-        # Model dimensions
-        self.k_fe = exog.shape[1] # Number of fixed effects parameters
-        if exog_re is not None:
-
+            if not self.data._param_names:
+                # HACK: could've been set in from_formula already
+                # needs refactor
+                (self.data.param_names,
+                 self.data.exog_re_names) = self._make_param_names(exog_re)
+            # Model dimensions
             # Number of random effect covariates
             self.k_re = exog_re.shape[1]
-
             # Number of covariance parameters
             self.k_re2 = self.k_re * (self.k_re + 1) // 2
 
-        else:
-            self.k_re = 1 # Default (random intercepts model)
-            self.k_re2 = 1
-
         # Override the default value
         self.nparams = self.k_fe + self.k_re2
 
-        self.data.xnames = self._make_param_names()
-
         # Convert the data to the internal representation, which is a
         # list of arrays, corresponding to the groups.
         group_labels = list(set(groups))
@@ -520,25 +529,23 @@ def __init__(self, endog, exog, groups, exog_re=None,
             self.exog_names = ["FE%d" % (k + 1) for k in
                                range(self.exog.shape[1])]
 
-        # Set the random effect parameter names
-        #if isinstance(self.exog_re, pd.DataFrame):
-        #    self.exog_re_names = list(self.exog_re.columns)
-        #else:
-        #    self.exog_re_names = ["Z%d" % (k+1) for k in
-        #                          range(self.exog_re.shape[1])]
+    def _make_param_names(self, exog_re):
+        exog_names = list(self.exog_names)
+        exog_re_names = _get_exog_re_names(exog_re)
+        param_names = []
 
-    def _make_param_names(self):
-        names = list(self.exog_names)
         jj = self.k_fe
-        for i in range(self.k_re):
+        for i in range(exog_re.shape[1]):
             for j in range(i + 1):
                 if i == j:
-                    names.append(self.exog_re_names[i] + " RE")
+                    param_names.append(exog_re_names[i] + " RE")
                 else:
-                    names.append(self.exog_re_names[j] + " x " +
-                                 self.exog_re_names[i] + " RE")
+                    param_names.append(exog_re_names[j] + " x " +
+                                       exog_re_names[i] + " RE")
                 jj += 1
 
+        return exog_names + exog_re_names, exog_re_names
+
     @classmethod
     def from_formula(cls, formula, data, re_formula=None, subset=None,
                      *args, **kwargs):
@@ -598,14 +605,15 @@ def from_formula(cls, formula, data, re_formula=None, subset=None,
         else:
             exog_re = np.ones((data.shape[0], 1),
                               dtype=np.float64)
-            exog_re_names = ["Intercept",]
+            exog_re_names = ["Intercept RE"]
 
         mod = super(MixedLM, cls).from_formula(formula, data,
                                                subset=None,
                                                exog_re=exog_re,
                                                *args, **kwargs)
 
-        mod.exog_re_names = exog_re_names
+        mod.data.param_names = mod.exog_names + exog_re_names
+        mod.data.exog_re_names = exog_re_names
 
         return mod
 
@@ -1740,6 +1748,7 @@ def __init__(self, model, params, cov_params):
         super(MixedLMResults, self).__init__(model, params,
            normalized_cov_params=cov_params)
 
+    @cache_readonly
     def bse_fe(self):
         """
         Returns the standard errors of the fixed effect regression
@@ -1748,6 +1757,7 @@ def bse_fe(self):
         p = self.model.exog.shape[1]
         return np.sqrt(np.diag(self.cov_params())[0:p])
 
+    @cache_readonly
     def bse_re(self):
         """
         Returns the standard errors of the variance parameters.  Note
@@ -1759,7 +1769,6 @@ def bse_re(self):
         p = self.model.exog.shape[1]
         return np.sqrt(self.scale * np.diag(self.cov_params())[p:])
 
-
     def ranef(self):
         """
         Returns the conditional means of all random effects given the
@@ -1893,7 +1902,6 @@ def summary(self, yname=None, xname_fe=None, xname_re=None,
 
         float_fmt = "%.3f"
 
-        names = list(self.model.exog_names)
         sdf = np.nan * np.ones((self.k_fe + self.k_re2, 6),
                                dtype=np.float64)
 
@@ -1919,17 +1927,11 @@ def summary(self, yname=None, xname_fe=None, xname_re=None,
         jj = self.k_fe
         for i in range(self.k_re):
             for j in range(i + 1):
-                import ipdb; ipdb.set_trace()
-                if i == j:
-                    names.append(self.model.exog_re_names[i] + " RE")
-                else:
-                    names.append(self.model.exog_re_names[j] + " x " +
-                                 self.model.exog_re_names[i] + " RE")
                 sdf[jj, 0] = self.cov_re[i, j]
                 sdf[jj, 1] = np.sqrt(self.scale) * self.bse[jj]
                 jj += 1
 
-        sdf = pd.DataFrame(index=names, data=sdf)
+        sdf = pd.DataFrame(index=self.model.data.param_names, data=sdf)
         sdf.columns = ['Coef.', 'Std.Err.', 'z', 'P>|z|',
                           '[' + str(alpha/2), str(1-alpha/2) + ']']
         for col in sdf.columns:
@@ -1940,7 +1942,6 @@ def summary(self, yname=None, xname_fe=None, xname_re=None,
 
         return smry
 
-
     def profile_re(self, re_ix, num_low=5, dist_low=1., num_high=5,
                    dist_high=1.):
         """
@@ -2031,4 +2032,8 @@ def profile_re(self, re_ix, num_low=5, dist_low=1., num_high=5,
 
 
 class MixedLMResultsWrapper(base.LikelihoodResultsWrapper):
-    pass
+    _attrs = {'bse_re': ('generic_columns', 'exog_re_names'),
+              'bse_fe': ('generic_columns', 'xnames'),
+              }
+    _upstream_attrs = base.LikelihoodResultsWrapper._wrap_attrs
+    _wrap_attrs = base.wrap.union_dicts(_attrs, _upstream_attrs)

From acfadf935f8030f8164168d1f48367e433952be3 Mon Sep 17 00:00:00 2001
From: Skipper Seabold <jsseabold@gmail.com>
Date: Thu, 25 Sep 2014 13:52:38 -0400
Subject: [PATCH 05/15] ENH: Handle the generic 2d case.

---
 statsmodels/base/data.py    | 16 ++++++++++++++--
 statsmodels/base/wrapper.py |  2 +-
 2 files changed, 15 insertions(+), 3 deletions(-)

diff --git a/statsmodels/base/data.py b/statsmodels/base/data.py
index 5c43347f3f5..3f4700dedf7 100644
--- a/statsmodels/base/data.py
+++ b/statsmodels/base/data.py
@@ -328,6 +328,8 @@ def wrap_output(self, obj, how='columns', names=None):
             return self.attach_cov_eq(obj)
         elif how == 'generic_columns':
             return self.attach_generic_columns(obj, names)
+        elif how == 'generic_columns_2d':
+            return self.attach_generic_columns_2d(obj, names)
         else:
             return obj
 
@@ -349,7 +351,10 @@ def attach_rows(self, result):
     def attach_dates(self, result):
         return result
 
-    def attach_generic_columns(self, result, names):
+    def attach_generic_columns(self, result, *args, **kwargs):
+        return result
+
+    def attach_generic_columns_2d(self, result, *args, **kwargs):
         return result
 
 
@@ -399,6 +404,12 @@ def attach_generic_columns(self, result, names):
         column_names = getattr(self, names, None)
         return Series(result, index=column_names)
 
+    def attach_generic_columns_2d(self, result, rownames, colnames=None):
+        colnames = colnames or rownames
+        rownames = getattr(self, rownames, None)
+        colnames = getattr(self, colnames, None)
+        return DataFrame(result, index=rownames, columns=colnames)
+
     def attach_columns(self, result):
         # this can either be a 1d array or a scalar
         # don't squeeze because it might be a 2d row array
@@ -412,7 +423,8 @@ def attach_columns_eq(self, result):
         return DataFrame(result, index=self.xnames, columns=self.ynames)
 
     def attach_cov(self, result):
-        return DataFrame(result, index=self.xnames, columns=self.xnames)
+        return DataFrame(result, index=self.param_names,
+                         columns=self.param_names)
 
     def attach_cov_eq(self, result):
         return DataFrame(result, index=self.ynames, columns=self.ynames)
diff --git a/statsmodels/base/wrapper.py b/statsmodels/base/wrapper.py
index c10246df170..ac25ee3f318 100644
--- a/statsmodels/base/wrapper.py
+++ b/statsmodels/base/wrapper.py
@@ -36,7 +36,7 @@ def __getattribute__(self, attr):
         data = results.model.data
         how = self._wrap_attrs.get(attr)
         if how and isinstance(how, tuple):
-            obj = data.wrap_output(obj, how=how[0], names=how[1])
+            obj = data.wrap_output(obj, how[0], *how[1:])
         elif how:
             obj = data.wrap_output(obj, how=how)
 

From 609a17e3a1aa0f00d5e9e0fe617a143b86c4ea6f Mon Sep 17 00:00:00 2001
From: Skipper Seabold <jsseabold@gmail.com>
Date: Thu, 25 Sep 2014 13:59:35 -0400
Subject: [PATCH 06/15] ENH: Wrap 2d attributes

---
 statsmodels/regression/mixed_linear_model.py | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/statsmodels/regression/mixed_linear_model.py b/statsmodels/regression/mixed_linear_model.py
index b91fc26393e..46760366b13 100644
--- a/statsmodels/regression/mixed_linear_model.py
+++ b/statsmodels/regression/mixed_linear_model.py
@@ -1769,6 +1769,7 @@ def bse_re(self):
         p = self.model.exog.shape[1]
         return np.sqrt(self.scale * np.diag(self.cov_params())[p:])
 
+    @cache_readonly
     def ranef(self):
         """
         Returns the conditional means of all random effects given the
@@ -1808,7 +1809,7 @@ def ranef(self):
 
         return ranef_dict
 
-
+    @cache_readonly
     def ranef_cov(self):
         """
         Returns the conditional covariance matrix of the random
@@ -2034,6 +2035,11 @@ def profile_re(self, re_ix, num_low=5, dist_low=1., num_high=5,
 class MixedLMResultsWrapper(base.LikelihoodResultsWrapper):
     _attrs = {'bse_re': ('generic_columns', 'exog_re_names'),
               'bse_fe': ('generic_columns', 'xnames'),
+              'cov_re': ('generic_columns_2d', 'exog_re_names'),
               }
     _upstream_attrs = base.LikelihoodResultsWrapper._wrap_attrs
     _wrap_attrs = base.wrap.union_dicts(_attrs, _upstream_attrs)
+
+    _methods = {}
+    _upstream_methods = base.LikelihoodResultsWrapper._wrap_methods
+    _wrap_methods = base.wrap.union_dicts(_methods, _upstream_methods)

From bfa893f9f47984a5a3841af78b6007ca84139881 Mon Sep 17 00:00:00 2001
From: Skipper Seabold <jsseabold@gmail.com>
Date: Thu, 25 Sep 2014 14:24:57 -0400
Subject: [PATCH 07/15] REF: Return DataFrame instead of dict.

---
 statsmodels/regression/mixed_linear_model.py | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)

diff --git a/statsmodels/regression/mixed_linear_model.py b/statsmodels/regression/mixed_linear_model.py
index 46760366b13..eb96b6c08a4 100644
--- a/statsmodels/regression/mixed_linear_model.py
+++ b/statsmodels/regression/mixed_linear_model.py
@@ -109,6 +109,8 @@
 from statsmodels.tools.sm_exceptions import ConvergenceWarning
 from statsmodels.base._penalties import Penalty
 
+from pandas import DataFrame
+
 
 def _get_exog_re_names(exog_re):
     if isinstance(exog_re, (pd.Series, pd.DataFrame)):
@@ -1782,7 +1784,6 @@ def ranef(self):
             variable to the conditional means of the random effects
             given the data.
         """
-
         try:
             cov_re_inv = np.linalg.inv(self.cov_re)
         except np.linalg.LinAlgError:
@@ -1807,7 +1808,10 @@ def ranef(self):
             ranef_dict[label] = np.dot(self.cov_re,
                                        np.dot(ex_r.T, vresid))
 
-        return ranef_dict
+        column_names = dict(zip(range(self.k_re),
+                                      self.model.data.exog_re_names))
+        df = DataFrame.from_dict(ranef_dict, orient='index')
+        return df.rename(columns=column_names).ix[self.model.group_labels]
 
     @cache_readonly
     def ranef_cov(self):
@@ -1829,10 +1833,9 @@ def ranef_cov(self):
             cov_re_inv = None
 
         ranef_dict = {}
+        #columns = self.model.data.exog_re_names
         for k in range(self.model.n_groups):
 
-            endog = self.model.endog_li[k]
-            exog = self.model.exog_li[k]
             ex_r = self.model.exog_re_li[k]
             ex2_r = self.model.exog_re2_li[k]
             label = self.model.group_labels[k]
@@ -1843,9 +1846,11 @@ def ranef_cov(self):
             mat2 = np.dot(mat1.T, mat2)
 
             ranef_dict[label] = self.cov_re - mat2
+            #ranef_dict[label] = DataFrame(self.cov_re - mat2,
+            #                              index=columns, columns=columns)
 
-        return ranef_dict
 
+        return ranef_dict
 
     def summary(self, yname=None, xname_fe=None, xname_re=None,
                 title=None, alpha=.05):

From 1f17d211cc802bd3b7c5a0e78ec2407e840585ad Mon Sep 17 00:00:00 2001
From: Skipper Seabold <jsseabold@gmail.com>
Date: Thu, 25 Sep 2014 14:34:47 -0400
Subject: [PATCH 08/15] TST: Fix tests for refactor

---
 statsmodels/regression/tests/test_lme.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/statsmodels/regression/tests/test_lme.py b/statsmodels/regression/tests/test_lme.py
index f955f78226c..cfdb19f4abd 100644
--- a/statsmodels/regression/tests/test_lme.py
+++ b/statsmodels/regression/tests/test_lme.py
@@ -298,9 +298,9 @@ def do1(self, reml, irf, ds_ix):
 
         # Not supported in R
         if not irf:
-            assert_almost_equal(mdf.ranef()[0], rslt.ranef_postmean,
+            assert_almost_equal(mdf.ranef.ix[0], rslt.ranef_postmean,
                                 decimal=3)
-            assert_almost_equal(mdf.ranef_cov()[0],
+            assert_almost_equal(mdf.ranef_cov[0],
                                 rslt.ranef_condvar,
                                 decimal=3)
 

From 0520e8987985e26f45b9156b95d9b7508074cb60 Mon Sep 17 00:00:00 2001
From: Skipper Seabold <jsseabold@gmail.com>
Date: Thu, 25 Sep 2014 17:13:23 -0400
Subject: [PATCH 09/15] STY: Rename attributes

---
 statsmodels/regression/mixed_linear_model.py | 21 ++++++++++----------
 statsmodels/regression/tests/test_lme.py     |  4 ++--
 2 files changed, 12 insertions(+), 13 deletions(-)

diff --git a/statsmodels/regression/mixed_linear_model.py b/statsmodels/regression/mixed_linear_model.py
index eb96b6c08a4..9669b593201 100644
--- a/statsmodels/regression/mixed_linear_model.py
+++ b/statsmodels/regression/mixed_linear_model.py
@@ -501,8 +501,7 @@ def __init__(self, endog, exog, groups, exog_re=None,
             # Number of covariance parameters
             self.k_re2 = self.k_re * (self.k_re + 1) // 2
 
-        # Override the default value
-        self.nparams = self.k_fe + self.k_re2
+        self.k_params = self.k_fe + self.k_re2
 
         # Convert the data to the internal representation, which is a
         # list of arrays, corresponding to the groups.
@@ -1751,7 +1750,7 @@ def __init__(self, model, params, cov_params):
            normalized_cov_params=cov_params)
 
     @cache_readonly
-    def bse_fe(self):
+    def standard_errors_fe(self):
         """
         Returns the standard errors of the fixed effect regression
         coefficients.
@@ -1760,7 +1759,7 @@ def bse_fe(self):
         return np.sqrt(np.diag(self.cov_params())[0:p])
 
     @cache_readonly
-    def bse_re(self):
+    def standard_errors_re(self):
         """
         Returns the standard errors of the variance parameters.  Note
         that the sampling distribution of variance parameters is
@@ -1772,17 +1771,17 @@ def bse_re(self):
         return np.sqrt(self.scale * np.diag(self.cov_params())[p:])
 
     @cache_readonly
-    def ranef(self):
+    def random_effects(self):
         """
         Returns the conditional means of all random effects given the
         data.
 
         Returns
         -------
-        ranef_dict : dict
-            A dictionary mapping the distinct values of the `group`
-            variable to the conditional means of the random effects
-            given the data.
+        random_effects : DataFrame
+            A DataFrame with the distinct `group` values as the index
+            and the conditional means of the random effects
+            in the columns.
         """
         try:
             cov_re_inv = np.linalg.inv(self.cov_re)
@@ -1814,14 +1813,14 @@ def ranef(self):
         return df.rename(columns=column_names).ix[self.model.group_labels]
 
     @cache_readonly
-    def ranef_cov(self):
+    def random_effects_cov(self):
         """
         Returns the conditional covariance matrix of the random
         effects for each group given the data.
 
         Returns
         -------
-        ranef_dict : dict
+        random_effects_cov : dict
             A dictionary mapping the distinct values of the `group`
             variable to the conditional covariance matrix of the
             random effects given the data.
diff --git a/statsmodels/regression/tests/test_lme.py b/statsmodels/regression/tests/test_lme.py
index cfdb19f4abd..679795a41fb 100644
--- a/statsmodels/regression/tests/test_lme.py
+++ b/statsmodels/regression/tests/test_lme.py
@@ -298,9 +298,9 @@ def do1(self, reml, irf, ds_ix):
 
         # Not supported in R
         if not irf:
-            assert_almost_equal(mdf.ranef.ix[0], rslt.ranef_postmean,
+            assert_almost_equal(mdf.random_effects.ix[0], rslt.ranef_postmean,
                                 decimal=3)
-            assert_almost_equal(mdf.ranef_cov[0],
+            assert_almost_equal(mdf.random_effects_cov[0],
                                 rslt.ranef_condvar,
                                 decimal=3)
 

From cba04face574d5b95c3bd89d31ba83bcf7781fed Mon Sep 17 00:00:00 2001
From: Skipper Seabold <jsseabold@gmail.com>
Date: Fri, 26 Sep 2014 00:36:23 -0400
Subject: [PATCH 10/15] DOC: Update release notes example

---
 docs/source/release/version0.6.rst | 13 +++++++------
 1 file changed, 7 insertions(+), 6 deletions(-)

diff --git a/docs/source/release/version0.6.rst b/docs/source/release/version0.6.rst
index 1e524f29a19..57c512f54fd 100644
--- a/docs/source/release/version0.6.rst
+++ b/docs/source/release/version0.6.rst
@@ -122,13 +122,14 @@ other types of random effects models can all be fit.
 Here is an example of fitting a random intercepts model to data from a
 longitudinal study:
 
-data = pd.read_csv("http://vincentarelbundock.github.io/Rdatasets/csv/geepack/dietox.csv")
-md = MixedLM.from_formula("Weight ~ Time", data, groups=data["Pig"])
-mdf = md.fit()
-print mdf.summary()
+.. ipython:: python
 
-To extend this to a random slopes model, we would add the statement
-`md.set_random("Time", data)` before calling the `fit` method.
+    import statsmodels.api as sm
+    import statsmodels.formula.api as smf
+    data = sm.datasets.get_rdataset('dietox', 'geepack', cache=True).data
+    md = smf.mixedlm("Weight ~ Time", data, groups=data["Pig"])
+    mdf = md.fit()
+    print mdf.summary()
 
 The Statsmodels LME framework currently supports post-estimation
 inference via Wald tests and confidence intervals on the coefficients,

From 29f0ecda16183ae5070feae985cc0cd0ea51a9e6 Mon Sep 17 00:00:00 2001
From: Skipper Seabold <jsseabold@gmail.com>
Date: Fri, 26 Sep 2014 09:20:25 -0400
Subject: [PATCH 11/15] ENH: Let t_test use param_names for better or worse.

---
 statsmodels/base/model.py | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/statsmodels/base/model.py b/statsmodels/base/model.py
index ab14b6e9662..4a568f1b5a4 100644
--- a/statsmodels/base/model.py
+++ b/statsmodels/base/model.py
@@ -1136,7 +1136,8 @@ def t_test(self, r_matrix, q_matrix=None, cov_p=None, scale=None,
                  "in 0.6.0. See the documentation for the new API",
                  FutureWarning)
             r_matrix = (r_matrix, q_matrix)
-        LC = DesignInfo(self.model.exog_names).linear_constraint(r_matrix)
+        names = self.model.data.param_names
+        LC = DesignInfo(names).linear_constraint(r_matrix)
         r_matrix, q_matrix = LC.coefs, LC.constants
         num_ttests = r_matrix.shape[0]
         num_params = r_matrix.shape[1]
@@ -1353,7 +1354,8 @@ def wald_test(self, r_matrix, q_matrix=None, cov_p=None, scale=1.0,
                  "in 0.6.0. See the documentation for the new API",
                  FutureWarning)
             r_matrix = (r_matrix, q_matrix)
-        LC = DesignInfo(self.model.exog_names).linear_constraint(r_matrix)
+        names = self.model.data.param_names
+        LC = DesignInfo(names).linear_constraint(r_matrix)
         r_matrix, q_matrix = LC.coefs, LC.constants
 
         if (self.normalized_cov_params is None and cov_p is None and

From a883cac78ee6cb40d20053af34bb69318b3db2aa Mon Sep 17 00:00:00 2001
From: Skipper Seabold <jsseabold@gmail.com>
Date: Fri, 26 Sep 2014 09:22:14 -0400
Subject: [PATCH 12/15] REF: Keep bse

---
 statsmodels/regression/mixed_linear_model.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/statsmodels/regression/mixed_linear_model.py b/statsmodels/regression/mixed_linear_model.py
index 9669b593201..5a4a2f193c0 100644
--- a/statsmodels/regression/mixed_linear_model.py
+++ b/statsmodels/regression/mixed_linear_model.py
@@ -1750,7 +1750,7 @@ def __init__(self, model, params, cov_params):
            normalized_cov_params=cov_params)
 
     @cache_readonly
-    def standard_errors_fe(self):
+    def bse_fe(self):
         """
         Returns the standard errors of the fixed effect regression
         coefficients.
@@ -1759,7 +1759,7 @@ def standard_errors_fe(self):
         return np.sqrt(np.diag(self.cov_params())[0:p])
 
     @cache_readonly
-    def standard_errors_re(self):
+    def bse_re(self):
         """
         Returns the standard errors of the variance parameters.  Note
         that the sampling distribution of variance parameters is

From c7bc0ff5c465298530c82011f2e3737a5a9147f5 Mon Sep 17 00:00:00 2001
From: Skipper Seabold <jsseabold@gmail.com>
Date: Fri, 26 Sep 2014 09:56:18 -0400
Subject: [PATCH 13/15] REF: likeval -> llf

---
 statsmodels/regression/mixed_linear_model.py | 6 ++----
 statsmodels/regression/tests/test_lme.py     | 2 +-
 2 files changed, 3 insertions(+), 5 deletions(-)

diff --git a/statsmodels/regression/mixed_linear_model.py b/statsmodels/regression/mixed_linear_model.py
index 5a4a2f193c0..75e52fbdd73 100644
--- a/statsmodels/regression/mixed_linear_model.py
+++ b/statsmodels/regression/mixed_linear_model.py
@@ -774,7 +774,6 @@ def fit_regularized(self, start_params=None, method='l1', alpha=0,
         results.method = mdf.method
         results.converged = True
         results.cov_pen = self.cov_pen
-        results.likeval = self.loglike(params_prof)
         results.k_fe = self.k_fe
         results.k_re = self.k_re
         results.k_re2 = self.k_re2
@@ -1701,7 +1700,6 @@ def fit(self, start_params=None, reml=True, niter_em=0,
         results.hist = hist
         results.reml = self.reml
         results.cov_pen = self.cov_pen
-        results.likeval = self.loglike(params)
         results.k_fe = self.k_fe
         results.k_re = self.k_re
         results.k_re2 = self.k_re2
@@ -1900,7 +1898,7 @@ def summary(self, yname=None, xname_fe=None, xname_re=None,
         info["Dependent Variable:"] = yname
         info["Method:"] = self.method
         info["Scale:"] = self.scale
-        info["Likelihood:"] = self.likeval
+        info["Likelihood:"] = self.llf
         info["Converged:"] = "Yes" if self.converged else "No"
         smry.add_dict(info)
         smry.add_title("Mixed Linear Model Regression Results")
@@ -2027,7 +2025,7 @@ def profile_re(self, re_ix, num_low=5, dist_low=1., num_high=5,
             params.set_cov_re(cov_re)
             rslt = model.fit(start_params=params, free=free,
                              reml=self.reml, cov_pen=self.cov_pen)
-            likev.append([rslt.cov_re[0, 0], rslt.likeval])
+            likev.append([rslt.cov_re[0, 0], rslt.llf])
         likev = np.asarray(likev)
 
         # Restore the original exog
diff --git a/statsmodels/regression/tests/test_lme.py b/statsmodels/regression/tests/test_lme.py
index 679795a41fb..298fb857192 100644
--- a/statsmodels/regression/tests/test_lme.py
+++ b/statsmodels/regression/tests/test_lme.py
@@ -294,7 +294,7 @@ def do1(self, reml, irf, ds_ix):
         assert_almost_equal(rslt.vcov_r, mdf.cov_params()[0:pf,0:pf],
                             decimal=3)
 
-        assert_almost_equal(mdf.likeval, rslt.loglike[0], decimal=2)
+        assert_almost_equal(mdf.llf, rslt.loglike[0], decimal=2)
 
         # Not supported in R
         if not irf:

From 76ba5ed4f4fe59eb873669c7af9d45416a7fb6d1 Mon Sep 17 00:00:00 2001
From: Skipper Seabold <jsseabold@gmail.com>
Date: Fri, 26 Sep 2014 10:17:34 -0400
Subject: [PATCH 14/15] ENH: Define df_resid so t_test works.

---
 statsmodels/regression/mixed_linear_model.py | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/statsmodels/regression/mixed_linear_model.py b/statsmodels/regression/mixed_linear_model.py
index 75e52fbdd73..554814d0c1f 100644
--- a/statsmodels/regression/mixed_linear_model.py
+++ b/statsmodels/regression/mixed_linear_model.py
@@ -108,6 +108,7 @@
 import warnings
 from statsmodels.tools.sm_exceptions import ConvergenceWarning
 from statsmodels.base._penalties import Penalty
+from statsmodels.compat.numpy import np_matrix_rank
 
 from pandas import DataFrame
 
@@ -524,6 +525,8 @@ def __init__(self, endog, exog, groups, exog_re=None,
 
         # The total number of observations, summed over all groups
         self.n_totobs = sum([len(y) for y in self.endog_li])
+        # why do it like the above?
+        self.nobs = len(self.endog)
 
         # Set the fixed effects parameter names
         if self.exog_names is None:
@@ -1745,7 +1748,9 @@ class MixedLMResults(base.LikelihoodModelResults):
     def __init__(self, model, params, cov_params):
 
         super(MixedLMResults, self).__init__(model, params,
-           normalized_cov_params=cov_params)
+                                             normalized_cov_params=cov_params)
+        self.nobs = self.model.nobs
+        self.df_resid = self.nobs - np_matrix_rank(self.model.exog)
 
     @cache_readonly
     def bse_fe(self):

From 4c7dc569a5e2eae28b1e2fec3aae2deed5113fff Mon Sep 17 00:00:00 2001
From: Skipper Seabold <jsseabold@gmail.com>
Date: Fri, 26 Sep 2014 10:49:53 -0400
Subject: [PATCH 15/15] TST: Mark slow test.

---
 statsmodels/regression/tests/test_lme.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/statsmodels/regression/tests/test_lme.py b/statsmodels/regression/tests/test_lme.py
index 298fb857192..9e7f39c790d 100644
--- a/statsmodels/regression/tests/test_lme.py
+++ b/statsmodels/regression/tests/test_lme.py
@@ -1,7 +1,8 @@
 import numpy as np
 import pandas as pd
 from statsmodels.regression.mixed_linear_model import MixedLM, MixedLMParams
-from numpy.testing import assert_almost_equal, assert_equal, assert_allclose
+from numpy.testing import (assert_almost_equal, assert_equal, assert_allclose,
+                           dec)
 from . import lme_r_results
 from statsmodels.base import _penalties as penalties
 import os
@@ -69,6 +70,7 @@ class TestMixedLM(object):
 
     # Test analytic scores using numeric differentiation
     # TODO: better checks on Hessian
+    @dec.slow
     def test_compare_numdiff(self):
 
         import statsmodels.tools.numdiff as nd