standardize signature for Index reductions, implement nanmean for datetime64 dtypes (#24293)

Author: jbrockmendel

pandas/core/base.py

@@ -973,10 +973,16 @@ def _ndarray_values(self):
     def empty(self):
         return not self.size
 
-    def max(self):
+    def max(self, axis=None, skipna=True):
         """
         Return the maximum value of the Index.
 
+        Parameters
+        ----------
+        axis : int, optional
+            For compatibility with NumPy. Only 0 or None are allowed.
+        skipna : bool, default True
+
         Returns
         -------
         scalar
@@ -1004,22 +1010,36 @@ def max(self):
         >>> idx.max()
         ('b', 2)
         """
-        return nanops.nanmax(self.values)
+        nv.validate_minmax_axis(axis)
+        return nanops.nanmax(self._values, skipna=skipna)
 
-    def argmax(self, axis=None):
+    def argmax(self, axis=None, skipna=True):
         """
         Return a ndarray of the maximum argument indexer.
 
+        Parameters
+        ----------
+        axis : {None}
+            Dummy argument for consistency with Series
+        skipna : bool, default True
+
         See Also
         --------
         numpy.ndarray.argmax
         """
-        return nanops.nanargmax(self.values)
+        nv.validate_minmax_axis(axis)
+        return nanops.nanargmax(self._values, skipna=skipna)
 
-    def min(self):
+    def min(self, axis=None, skipna=True):
         """
         Return the minimum value of the Index.
 
+        Parameters
+        ----------
+        axis : {None}
+            Dummy argument for consistency with Series
+        skipna : bool, default True
+
         Returns
         -------
         scalar
@@ -1047,17 +1067,25 @@ def min(self):
         >>> idx.min()
         ('a', 1)
         """
-        return nanops.nanmin(self.values)
+        nv.validate_minmax_axis(axis)
+        return nanops.nanmin(self._values, skipna=skipna)
 
-    def argmin(self, axis=None):
+    def argmin(self, axis=None, skipna=True):
         """
         Return a ndarray of the minimum argument indexer.
 
+        Parameters
+        ----------
+        axis : {None}
+            Dummy argument for consistency with Series
+        skipna : bool, default True
+
         See Also
         --------
         numpy.ndarray.argmin
         """
-        return nanops.nanargmin(self.values)
+        nv.validate_minmax_axis(axis)
+        return nanops.nanargmin(self._values, skipna=skipna)
 
     def tolist(self):
         """
@@ -1110,7 +1138,7 @@ def _reduce(self, op, name, axis=0, skipna=True, numeric_only=None,
         if func is None:
             raise TypeError("{klass} cannot perform the operation {op}".format(
                             klass=self.__class__.__name__, op=name))
-        return func(**kwds)
+        return func(skipna=skipna, **kwds)
 
     def _map_values(self, mapper, na_action=None):
         """

pandas/core/dtypes/missing.py

@@ -198,6 +198,8 @@ def _isna_ndarraylike(obj):
         else:
             values = obj
         result = values.isna()
+    elif isinstance(obj, ABCDatetimeArray):
+        return obj.isna()
     elif is_string_dtype(dtype):
         # Working around NumPy ticket 1542
         shape = values.shape

pandas/core/indexes/datetimelike.py

@@ -267,35 +267,41 @@ def tolist(self):
         """
         return list(self.astype(object))
 
-    def min(self, axis=None, *args, **kwargs):
+    def min(self, axis=None, skipna=True, *args, **kwargs):
         """
         Return the minimum value of the Index or minimum along
         an axis.
 
         See Also
         --------
         numpy.ndarray.min
+        Series.min : Return the minimum value in a Series.
         """
         nv.validate_min(args, kwargs)
         nv.validate_minmax_axis(axis)
 
-        try:
-            i8 = self.asi8
+        if not len(self):
+            return self._na_value
 
+        i8 = self.asi8
+        try:
             # quick check
             if len(i8) and self.is_monotonic:
                 if i8[0] != iNaT:
                     return self._box_func(i8[0])
 
             if self.hasnans:
-                min_stamp = self[~self._isnan].asi8.min()
+                if skipna:
+                    min_stamp = self[~self._isnan].asi8.min()
+                else:
+                    return self._na_value
             else:
                 min_stamp = i8.min()
             return self._box_func(min_stamp)
         except ValueError:
             return self._na_value
 
-    def argmin(self, axis=None, *args, **kwargs):
+    def argmin(self, axis=None, skipna=True, *args, **kwargs):
         """
         Returns the indices of the minimum values along an axis.
 
@@ -312,41 +318,47 @@ def argmin(self, axis=None, *args, **kwargs):
         i8 = self.asi8
         if self.hasnans:
             mask = self._isnan
-            if mask.all():
+            if mask.all() or not skipna:
                 return -1
             i8 = i8.copy()
             i8[mask] = np.iinfo('int64').max
         return i8.argmin()
 
-    def max(self, axis=None, *args, **kwargs):
+    def max(self, axis=None, skipna=True, *args, **kwargs):
         """
         Return the maximum value of the Index or maximum along
         an axis.
 
         See Also
         --------
         numpy.ndarray.max
+        Series.max : Return the maximum value in a Series.
         """
         nv.validate_max(args, kwargs)
         nv.validate_minmax_axis(axis)
 
-        try:
-            i8 = self.asi8
+        if not len(self):
+            return self._na_value
 
+        i8 = self.asi8
+        try:
             # quick check
             if len(i8) and self.is_monotonic:
                 if i8[-1] != iNaT:
                     return self._box_func(i8[-1])
 
             if self.hasnans:
-                max_stamp = self[~self._isnan].asi8.max()
+                if skipna:
+                    max_stamp = self[~self._isnan].asi8.max()
+                else:
+                    return self._na_value
             else:
                 max_stamp = i8.max()
             return self._box_func(max_stamp)
         except ValueError:
             return self._na_value
 
-    def argmax(self, axis=None, *args, **kwargs):
+    def argmax(self, axis=None, skipna=True, *args, **kwargs):
         """
         Returns the indices of the maximum values along an axis.
 
@@ -363,7 +375,7 @@ def argmax(self, axis=None, *args, **kwargs):
         i8 = self.asi8
         if self.hasnans:
             mask = self._isnan
-            if mask.all():
+            if mask.all() or not skipna:
                 return -1
             i8 = i8.copy()
             i8[mask] = 0

pandas/core/indexes/range.py

@@ -297,12 +297,14 @@ def _minmax(self, meth):
 
         return self._start + self._step * no_steps
 
-    def min(self):
+    def min(self, axis=None, skipna=True):
         """The minimum value of the RangeIndex"""
+        nv.validate_minmax_axis(axis)
         return self._minmax('min')
 
-    def max(self):
+    def max(self, axis=None, skipna=True):
         """The maximum value of the RangeIndex"""
+        nv.validate_minmax_axis(axis)
         return self._minmax('max')
 
     def argsort(self, *args, **kwargs):

pandas/core/nanops.py

@@ -6,14 +6,14 @@
 
 import numpy as np
 
-from pandas._libs import lib, tslibs
+from pandas._libs import iNaT, lib, tslibs
 import pandas.compat as compat
 
 from pandas.core.dtypes.cast import _int64_max, maybe_upcast_putmask
 from pandas.core.dtypes.common import (
     _get_dtype, is_any_int_dtype, is_bool_dtype, is_complex, is_complex_dtype,
-    is_datetime64_dtype, is_datetime_or_timedelta_dtype, is_float,
-    is_float_dtype, is_integer, is_integer_dtype, is_numeric_dtype,
+    is_datetime64_dtype, is_datetime64tz_dtype, is_datetime_or_timedelta_dtype,
+    is_float, is_float_dtype, is_integer, is_integer_dtype, is_numeric_dtype,
     is_object_dtype, is_scalar, is_timedelta64_dtype)
 from pandas.core.dtypes.missing import isna, na_value_for_dtype, notna
 
@@ -203,15 +203,28 @@ def _get_values(values, skipna, fill_value=None, fill_value_typ=None,
     if necessary copy and mask using the specified fill_value
     copy = True will force the copy
     """
-    values = com.values_from_object(values)
+
+    if is_datetime64tz_dtype(values):
+        # com.values_from_object returns M8[ns] dtype instead of tz-aware,
+        #  so this case must be handled separately from the rest
+        dtype = values.dtype
+        values = getattr(values, "_values", values)
+    else:
+        values = com.values_from_object(values)
+        dtype = values.dtype
 
     if mask is None:
         if isfinite:
             mask = _isfinite(values)
         else:
             mask = isna(values)
 
-    dtype = values.dtype
+    if is_datetime_or_timedelta_dtype(values) or is_datetime64tz_dtype(values):
+        # changing timedelta64/datetime64 to int64 needs to happen after
+        #  finding `mask` above
+        values = getattr(values, "asi8", values)
+        values = values.view(np.int64)
+
     dtype_ok = _na_ok_dtype(dtype)
 
     # get our fill value (in case we need to provide an alternative
@@ -232,8 +245,6 @@ def _get_values(values, skipna, fill_value=None, fill_value_typ=None,
     elif copy:
         values = values.copy()
 
-    values = _view_if_needed(values)
-
     # return a platform independent precision dtype
     dtype_max = dtype
     if is_integer_dtype(dtype) or is_bool_dtype(dtype):
@@ -259,21 +270,19 @@ def _na_ok_dtype(dtype):
                           (np.integer, np.timedelta64, np.datetime64))
 
 
-def _view_if_needed(values):
-    if is_datetime_or_timedelta_dtype(values):
-        return values.view(np.int64)
-    return values
-
-
 def _wrap_results(result, dtype, fill_value=None):
     """ wrap our results if needed """
 
-    if is_datetime64_dtype(dtype):
+    if is_datetime64_dtype(dtype) or is_datetime64tz_dtype(dtype):
+        if fill_value is None:
+            # GH#24293
+            fill_value = iNaT
         if not isinstance(result, np.ndarray):
+            tz = getattr(dtype, 'tz', None)
             assert not isna(fill_value), "Expected non-null fill_value"
             if result == fill_value:
                 result = np.nan
-            result = tslibs.Timestamp(result)
+            result = tslibs.Timestamp(result, tz=tz)
         else:
             result = result.view(dtype)
     elif is_timedelta64_dtype(dtype):
@@ -426,7 +435,6 @@ def nansum(values, axis=None, skipna=True, min_count=0, mask=None):
     return _wrap_results(the_sum, dtype)
 
 
-@disallow('M8')
 @bottleneck_switch()
 def nanmean(values, axis=None, skipna=True, mask=None):
     """
@@ -457,7 +465,8 @@ def nanmean(values, axis=None, skipna=True, mask=None):
         values, skipna, 0, mask=mask)
     dtype_sum = dtype_max
     dtype_count = np.float64
-    if is_integer_dtype(dtype) or is_timedelta64_dtype(dtype):
+    if (is_integer_dtype(dtype) or is_timedelta64_dtype(dtype) or
+            is_datetime64_dtype(dtype) or is_datetime64tz_dtype(dtype)):
         dtype_sum = np.float64
     elif is_float_dtype(dtype):
         dtype_sum = dtype
@@ -466,7 +475,9 @@ def nanmean(values, axis=None, skipna=True, mask=None):
     the_sum = _ensure_numeric(values.sum(axis, dtype=dtype_sum))
 
     if axis is not None and getattr(the_sum, 'ndim', False):
-        the_mean = the_sum / count
+        with np.errstate(all="ignore"):
+            # suppress division by zero warnings
+            the_mean = the_sum / count
         ct_mask = count == 0
         if ct_mask.any():
             the_mean[ct_mask] = np.nan

pandas/core/series.py

@@ -17,9 +17,10 @@
 
 from pandas.core.dtypes.common import (
     _is_unorderable_exception, ensure_platform_int, is_bool,
-    is_categorical_dtype, is_datetime64tz_dtype, is_datetimelike, is_dict_like,
-    is_extension_array_dtype, is_extension_type, is_hashable, is_integer,
-    is_iterator, is_list_like, is_scalar, is_string_like, is_timedelta64_dtype)
+    is_categorical_dtype, is_datetime64_dtype, is_datetime64tz_dtype,
+    is_datetimelike, is_dict_like, is_extension_array_dtype, is_extension_type,
+    is_hashable, is_integer, is_iterator, is_list_like, is_scalar,
+    is_string_like, is_timedelta64_dtype)
 from pandas.core.dtypes.generic import (
     ABCDataFrame, ABCDatetimeIndex, ABCSeries, ABCSparseArray, ABCSparseSeries)
 from pandas.core.dtypes.missing import (
@@ -3537,6 +3538,9 @@ def _reduce(self, op, name, axis=0, skipna=True, numeric_only=None,
         # dispatch to ExtensionArray interface
         if isinstance(delegate, ExtensionArray):
             return delegate._reduce(name, skipna=skipna, **kwds)
+        elif is_datetime64_dtype(delegate):
+            # use DatetimeIndex implementation to handle skipna correctly
+            delegate = DatetimeIndex(delegate)
 
         # dispatch to numpy arrays
         elif isinstance(delegate, np.ndarray):

pandas/tests/indexes/test_range.py

@@ -895,10 +895,18 @@ def test_max_min(self, start, stop, step):
         result = idx.max()
         assert result == expected
 
+        # skipna should be irrelevant since RangeIndex should never have NAs
+        result2 = idx.max(skipna=False)
+        assert result2 == expected
+
         expected = idx._int64index.min()
         result = idx.min()
         assert result == expected
 
+        # skipna should be irrelevant since RangeIndex should never have NAs
+        result2 = idx.min(skipna=False)
+        assert result2 == expected
+
         # empty
         idx = RangeIndex(start, stop, -step)
         assert isna(idx.max())