ENH: add fill_value to asfreq, see #3715

Author: nchmura4

doc/source/whatsnew/v0.20.0.txt

@@ -111,9 +111,8 @@ Other enhancements
 - ``pd.DataFrame.plot`` now prints a title above each subplot if ``suplots=True`` and ``title`` is a list of strings (:issue:`14753`)
 - ``pd.Series.interpolate`` now supports timedelta as an index type with ``method='time'`` (:issue:`6424`)
 - ``pandas.io.json.json_normalize()`` gained the option ``errors='ignore'|'raise'``; the default is ``errors='raise'`` which is backward compatible. (:issue:`14583`)
-
 - ``.select_dtypes()`` now allows the string 'datetimetz' to generically select datetimes with tz (:issue:`14910`)
-
+- ``DataFrame.asfreq()`` now accepts a ``fill_value`` option to fill missing values during resampling (:issue:`3715`).
 
 .. _whatsnew_0200.api_breaking:
 

pandas/core/generic.py

@@ -4054,12 +4054,17 @@ def groupby(self, by=None, axis=0, level=None, as_index=True, sort=True,
                        sort=sort, group_keys=group_keys, squeeze=squeeze,
                        **kwargs)
 
-    def asfreq(self, freq, method=None, how=None, normalize=False):
+    def asfreq(self, freq, method=None, how=None, normalize=False,
+               fill_value=None):
         """
         Convert TimeSeries to specified frequency.
 
         Optionally provide filling method to pad/backfill missing values.
 
+        Returns the original data conformed to a new index with the specified
+        frequency. ``resample`` is more appropriate if an operation, such as
+        summarization, is necessary to represent the data at the new frequency.
+
         Parameters
         ----------
         freq : DateOffset object, or string
@@ -4074,18 +4079,79 @@ def asfreq(self, freq, method=None, how=None, normalize=False):
             For PeriodIndex only, see PeriodIndex.asfreq
         normalize : bool, default False
             Whether to reset output index to midnight
+        fill_value: scalar, optional
+            Value to use for missing values, applied during upsampling (note
+            this does not fill NaNs that already were present).
+
+            .. versionadded:: 0.20.0
 
         Returns
         -------
         converted : type of caller
 
+        Examples
+        --------
+
+        Start by creating a series with 4 one minute timestamps.
+
+        >>> index = pd.date_range('1/1/2000', periods=4, freq='T')
+        >>> series = pd.Series([0.0, None, 2.0, 3.0], index=index)
+        >>> df = pd.DataFrame({'s':series})
+        >>> df
+                               s
+        2000-01-01 00:00:00    0.0
+        2000-01-01 00:01:00    NaN
+        2000-01-01 00:02:00    2.0
+        2000-01-01 00:03:00    3.0
+
+        Upsample the series into 30 second bins.
+
+        >>> df.asfreq(freq='30S')
+                               s
+        2000-01-01 00:00:00    0.0
+        2000-01-01 00:00:30    NaN
+        2000-01-01 00:01:00    NaN
+        2000-01-01 00:01:30    NaN
+        2000-01-01 00:02:00    2.0
+        2000-01-01 00:02:30    NaN
+        2000-01-01 00:03:00    3.0
+
+        Upsample again, providing a ``fill value``.
+
+        >>> df.asfreq(freq='30S', fill_value=9.0)
+                               s
+        2000-01-01 00:00:00    0.0
+        2000-01-01 00:00:30    9.0
+        2000-01-01 00:01:00    NaN
+        2000-01-01 00:01:30    9.0
+        2000-01-01 00:02:00    2.0
+        2000-01-01 00:02:30    9.0
+        2000-01-01 00:03:00    3.0
+
+        Upsample again, providing a ``method``.
+
+        >>> df.asfreq(freq='30S', method='bfill')
+                               s
+        2000-01-01 00:00:00    0.0
+        2000-01-01 00:00:30    NaN
+        2000-01-01 00:01:00    NaN
+        2000-01-01 00:01:30    2.0
+        2000-01-01 00:02:00    2.0
+        2000-01-01 00:02:30    3.0
+        2000-01-01 00:03:00    3.0
+
+        See Also
+        --------
+        reindex
+
         Notes
         -----
         To learn more about the frequency strings, please see `this link
         <http://pandas.pydata.org/pandas-docs/stable/timeseries.html#offset-aliases>`__.
         """
         from pandas.tseries.resample import asfreq
-        return asfreq(self, freq, method=method, how=how, normalize=normalize)
+        return asfreq(self, freq, method=method, how=how, normalize=normalize,
+                      fill_value=fill_value)
 
     def at_time(self, time, asof=False):
         """
@@ -4166,9 +4232,6 @@ def resample(self, rule, how=None, axis=0, fill_method=None, closed=None,
 
             .. versionadded:: 0.19.0
 
-        Notes
-        -----
-
         To learn more about the offset strings, please see `this link
         <http://pandas.pydata.org/pandas-docs/stable/timeseries.html#offset-aliases>`__.
 

pandas/tests/frame/test_timeseries.py

@@ -323,6 +323,26 @@ def test_asfreq_datetimeindex(self):
         ts = df['A'].asfreq('B')
         tm.assertIsInstance(ts.index, DatetimeIndex)
 
+    def test_asfreq_fillvalue(self):
+        # test for fill value during upsampling, related to issue 3715
+
+        # setup
+        rng = pd.date_range('1/1/2016', periods=10, freq='2S')
+        ts = pd.Series(np.arange(len(rng)), index=rng)
+        df = pd.DataFrame({'one': ts})
+
+        # insert pre-existing missing value
+        df.loc['2016-01-01 00:00:08', 'one'] = None
+
+        actual_df = df.asfreq(freq='1S', fill_value=9.0)
+        expected_df = df.asfreq(freq='1S').fillna(9.0)
+        expected_df.loc['2016-01-01 00:00:08', 'one'] = None
+        assert_frame_equal(expected_df, actual_df)
+
+        expected_series = ts.asfreq(freq='1S').fillna(9.0)
+        actual_series = ts.asfreq(freq='1S', fill_value=9.0)
+        assert_series_equal(expected_series, actual_series)
+
     def test_first_last_valid(self):
         N = len(self.frame.index)
         mat = randn(N)

pandas/tseries/resample.py

@@ -352,7 +352,7 @@ def transform(self, arg, *args, **kwargs):
     def _downsample(self, f):
         raise AbstractMethodError(self)
 
-    def _upsample(self, f, limit=None):
+    def _upsample(self, f, limit=None, fill_value=None):
         raise AbstractMethodError(self)
 
     def _gotitem(self, key, ndim, subset=None):
@@ -499,12 +499,25 @@ def interpolate(self, method='linear', axis=0, limit=None, inplace=False,
                                   limit_direction=limit_direction,
                                   downcast=downcast, **kwargs)
 
-    def asfreq(self):
+    def asfreq(self, fill_value=None):
         """
         return the values at the new freq,
-        essentially a reindex with (no filling)
+        essentially a reindex
+
+        Parameters
+        ----------
+        fill_value: scalar, optional
+            Value to use for missing values, applied during upsampling (note
+            this does not fill NaNs that already were present).
+
+            .. versionadded:: 0.20.0
+
+        See Also
+        --------
+        Series.asfreq
+        DataFrame.asfreq
         """
-        return self._upsample('asfreq')
+        return self._upsample('asfreq', fill_value=fill_value)
 
     def std(self, ddof=1, *args, **kwargs):
         """
@@ -704,12 +717,14 @@ def _adjust_binner_for_upsample(self, binner):
             binner = binner[:-1]
         return binner
 
-    def _upsample(self, method, limit=None):
+    def _upsample(self, method, limit=None, fill_value=None):
         """
         method : string {'backfill', 'bfill', 'pad',
             'ffill', 'asfreq'} method for upsampling
         limit : int, default None
             Maximum size gap to fill when reindexing
+        fill_value : scalar, default None
+            Value to use for missing values
 
         See also
         --------
@@ -736,7 +751,7 @@ def _upsample(self, method, limit=None):
             result.index = res_index
         else:
             result = obj.reindex(res_index, method=method,
-                                 limit=limit)
+                                 limit=limit, fill_value=fill_value)
 
         return self._wrap_result(result)
 
@@ -855,12 +870,14 @@ def _downsample(self, how, **kwargs):
             'Frequency {} cannot be resampled to {}, as they are not '
             'sub or super periods'.format(ax.freq, self.freq))
 
-    def _upsample(self, method, limit=None):
+    def _upsample(self, method, limit=None, fill_value=None):
         """
         method : string {'backfill', 'bfill', 'pad', 'ffill'}
             method for upsampling
         limit : int, default None
             Maximum size gap to fill when reindexing
+        fill_value : scalar, default None
+            Value to use for missing values
 
         See also
         --------
@@ -874,8 +891,8 @@ def _upsample(self, method, limit=None):
                              " datetime-like")
         # we may need to actually resample as if we are timestamps
         if self.kind == 'timestamp':
-            return super(PeriodIndexResampler, self)._upsample(method,
-                                                               limit=limit)
+            return super(PeriodIndexResampler, self)._upsample(
+                method, limit=limit, fill_value=fill_value)
 
         ax = self.ax
         obj = self.obj
@@ -889,7 +906,6 @@ def _upsample(self, method, limit=None):
         return self._wrap_result(_take_new_index(
             obj, indexer, new_index, axis=self.axis))
 
-
 class PeriodIndexResamplerGroupby(_GroupByMixin, PeriodIndexResampler):
     """
     Provides a resample of a groupby implementation
@@ -1333,7 +1349,7 @@ def _adjust_dates_anchored(first, last, offset, closed='right', base=0):
             Timestamp(lresult).tz_localize(last_tzinfo, ambiguous=last_dst))
 
 
-def asfreq(obj, freq, method=None, how=None, normalize=False):
+def asfreq(obj, freq, method=None, how=None, normalize=False, fill_value=None):
     """
     Utility frequency conversion method for Series/DataFrame
     """
@@ -1353,7 +1369,7 @@ def asfreq(obj, freq, method=None, how=None, normalize=False):
             return obj.copy()
         dti = date_range(obj.index[0], obj.index[-1], freq=freq)
         dti.name = obj.index.name
-        rs = obj.reindex(dti, method=method)
+        rs = obj.reindex(dti, method=method, fill_value=fill_value)
         if normalize:
             rs.index = rs.index.normalize()
         return rs

pandas/tseries/tests/test_resample.py

@@ -693,6 +693,24 @@ def test_asfreq_upsample(self):
         expected = frame.reindex(new_index)
         assert_frame_equal(result, expected)
 
+    def test_asfreq_fill_value(self):
+        # test for fill value during resampling, issue 3715
+
+        s = self.create_series()
+
+        result = s.resample('1H').asfreq()
+        new_index = self.create_index(s.index[0], s.index[-1], freq='1H')
+        expected = s.reindex(new_index)
+        assert_series_equal(result, expected)
+
+        frame = s.to_frame('value')
+        frame.iloc[1] = None
+        result = frame.resample('1H').asfreq(fill_value=4.0)
+        new_index = self.create_index(frame.index[0],
+                                      frame.index[-1], freq='1H')
+        expected = frame.reindex(new_index, fill_value=4.0)
+        assert_frame_equal(result, expected)
+
     def test_resample_interpolate(self):
         # # 12925
         df = self.create_series().to_frame('value')
@@ -2122,6 +2140,25 @@ def test_asfreq_upsample(self):
         result = frame.resample('1H').asfreq()
         assert_frame_equal(result, expected)
 
+    def test_asfreq_fill_value(self):
+        # test for fill value during resampling, issue 3715
+
+        s = self.create_series()
+        new_index = date_range(s.index[0].to_timestamp(how='start'),
+                               (s.index[-1]).to_timestamp(how='start'),
+                               freq='1H')
+        expected = s.to_timestamp().reindex(new_index, fill_value=4.0)
+        result = s.resample('1H', kind='timestamp').asfreq(fill_value=4.0)
+        assert_series_equal(result, expected)
+
+        frame = s.to_frame('value')
+        new_index = date_range(frame.index[0].to_timestamp(how='start'),
+                               (frame.index[-1]).to_timestamp(how='start'),
+                               freq='1H')
+        expected = frame.to_timestamp().reindex(new_index, fill_value=3.0)
+        result = frame.resample('1H', kind='timestamp').asfreq(fill_value=3.0)
+        assert_frame_equal(result, expected)
+
     def test_selection(self):
         index = self.create_series().index
         # This is a bug, these should be implemented