COMPAT: Pandas 0.20 compat

The tests all pass on the RC, so we don't have any substantial changes.

This fixes some new FutureWarnings, related to pandas new module privacy.

Closes #3617

Author: TomAugspurger

statsmodels/compat/pandas.py

@@ -55,10 +55,10 @@ def is_numeric_dtype(arr_or_dtype):
             return (issubclass(tipo, (np.number, np.bool_)) and
                     not issubclass(tipo, (np.datetime64, np.timedelta64)))
 
-try:
-    import pandas.tseries.tools as datetools
-    import pandas.tseries.frequencies as frequencies
-except ImportError:
-    from pandas.core import datetools
-    frequencies = datetools
-
+if version >= '0.20':
+    from pandas.tseries import frequencies
+else:
+    try:
+        import pandas.tseries.frequencies as frequencies
+    except ImportError:
+        from pandas.core import datetools as frequencies

statsmodels/discrete/discrete_model.py

@@ -38,7 +38,7 @@
 import statsmodels.regression.linear_model as lm
 import statsmodels.base.wrapper as wrap
 from statsmodels.compat.numpy import np_matrix_rank
-from pandas.core.api import get_dummies
+from pandas import get_dummies
 
 from statsmodels.base.l1_slsqp import fit_l1_slsqp
 try:

statsmodels/graphics/tests/test_tsaplots.py

@@ -1,5 +1,4 @@
-from statsmodels.compat.python import lmap, map
-from statsmodels.compat.pandas import datetools
+from statsmodels.compat.python import lmap
 import numpy as np
 import pandas as pd
 from numpy.testing import dec, assert_equal
@@ -92,8 +91,7 @@ def test_plot_month():
     dta = sm.datasets.elnino.load_pandas().data
     dta['YEAR'] = dta.YEAR.astype(int).apply(str)
     dta = dta.set_index('YEAR').T.unstack()
-    dates = lmap(lambda x: datetools.parse_time_string('1 '+' '.join(x))[0],
-                 dta.index.values)
+    dates = pd.to_datetime(['-'.join([x[1], x[0]]) for x in dta.index.values])
 
     # test dates argument
     fig = month_plot(dta.values, dates=dates, ylabel='el nino')
@@ -118,28 +116,24 @@ def test_plot_month():
 def test_plot_quarter():
     dta = sm.datasets.macrodata.load_pandas().data
     dates = lmap('Q'.join, zip(dta.year.astype(int).apply(str),
-                              dta.quarter.astype(int).apply(str)))
+                               dta.quarter.astype(int).apply(str)))
     # test dates argument
     quarter_plot(dta.unemp.values, dates)
     plt.close('all')
 
     # test with a DatetimeIndex with no freq
-    parser = datetools.parse_time_string
-    dta.set_index(pd.DatetimeIndex((x[0] for x in map(parser, dates))),
-                  inplace=True)
+    dta.set_index(pd.to_datetime(dates), inplace=True)
     quarter_plot(dta.unemp)
     plt.close('all')
 
     # w freq
     # see pandas #6631
-    dta.index = pd.DatetimeIndex((x[0] for x in map(parser, dates)),
-                                   freq='QS-Oct')
+    dta.index = pd.DatetimeIndex(pd.to_datetime(dates), freq='QS-Oct')
     quarter_plot(dta.unemp)
     plt.close('all')
 
     # w PeriodIndex
-    dta.index = pd.PeriodIndex((x[0] for x in map(parser, dates)),
-                                   freq='Q')
+    dta.index = pd.PeriodIndex(pd.to_datetime(dates), freq='Q')
     quarter_plot(dta.unemp)
     plt.close('all')
 

statsmodels/tools/grouputils.py

@@ -34,7 +34,7 @@
 from statsmodels.compat.numpy import npc_unique
 from statsmodels.compat.pandas import sort_values
 import statsmodels.tools.data as data_util
-from pandas.core.index import Index, MultiIndex
+from pandas import Index, MultiIndex
 
 
 def combine_indices(groups, prefix='', sep='.', return_labels=False):

statsmodels/tsa/base/datetools.py

@@ -3,11 +3,11 @@
 """
 from statsmodels.compat.python import (lrange, lzip, lmap, string_types, long,
                                        callable, asstr, reduce, zip, map)
-from statsmodels.compat.pandas import datetools
 import re
 import datetime
 
-from pandas import Int64Index, Period, PeriodIndex, Timestamp, DatetimeIndex
+from pandas import (Int64Index, Period, PeriodIndex, Timestamp, DatetimeIndex,
+                    to_datetime)
 import numpy as np
 
 _quarter_to_day = {
@@ -82,7 +82,7 @@ def date_parser(timestr, parserinfo=None, **kwargs):
         month, day = 12, 31
         year = int(timestr)
     else:
-        return datetools.to_datetime(timestr, **kwargs)
+        return to_datetime(timestr, **kwargs)
 
     return datetime.datetime(year, month, day)
 
@@ -176,7 +176,8 @@ def dates_from_range(start, end=None, length=None):
     Examples
     --------
     >>> import statsmodels.api as sm
-    >>> dates = sm.tsa.datetools.date_range('1960m1', length=nobs)
+    >>> import pandas as pd
+    >>> dates = pd.date_range('1960m1', length=nobs)
 
 
     Returns

statsmodels/tsa/vector_ar/util.py

@@ -110,8 +110,6 @@ def parse_lutkepohl_data(path): # pragma: no cover
     Source for data files: www.jmulti.de
     """
 
-    from statsmodels.compat.pandas import datetools as dt
-
     from collections import deque
     from datetime import datetime
     import pandas