def normalize_keyword_aggregation(
kwargs: dict) -> Tuple[dict, List[str], List[int]]:
"""
Normalize user-provided "named aggregation" kwargs.
Transforms from the new ``Mapping[str, NamedAgg]`` style kwargs
to the old Dict[str, List[scalar]]].
Parameters
----------
kwargs : dict
Returns
-------
aggspec : dict
The transformed kwargs.
columns : List[str]
The user-provided keys.
col_idx_order : List[int]
List of columns indices.
Examples
--------
>>> normalize_keyword_aggregation({"output": ("input", "sum")})
(defaultdict(<class 'list'>, {'input': ['sum']}), ('output',), array([0]))
"""
# Normalize the aggregation functions as Mapping[column, List[func]],
# process normally, then fixup the names.
# TODO: aggspec type: typing.Dict[str, List[AggScalar]]
# May be hitting https://github.com/python/mypy/issues/5958
# saying it doesn't have an attribute __name__
aggspec: DefaultDict = defaultdict(list)
order = []
columns, pairs = list(zip(*kwargs.items()))
for name, (column, aggfunc) in zip(columns, pairs):
aggspec[column].append(aggfunc)
order.append((column, com.get_callable_name(aggfunc) or aggfunc))
# uniquify aggfunc name if duplicated in order list
uniquified_order = _make_unique_kwarg_list(order)
# GH 25719, due to aggspec will change the order of assigned columns in aggregation
# uniquified_aggspec will store uniquified order list and will compare it with order
# based on index
aggspec_order = [(column, com.get_callable_name(aggfunc) or aggfunc)
for column, aggfuncs in aggspec.items()
for aggfunc in aggfuncs]
uniquified_aggspec = _make_unique_kwarg_list(aggspec_order)
# get the new indice of columns by comparison
col_idx_order = Index(uniquified_aggspec).get_indexer(uniquified_order)
return aggspec, columns, col_idx_order
def _managle_lambda_list(aggfuncs: Sequence[Any]) -> Sequence[Any]:
"""
Possibly mangle a list of aggfuncs.
Parameters
----------
aggfuncs : Sequence
Returns
-------
mangled: list-like
A new AggSpec sequence, where lambdas have been converted
to have unique names.
Notes
-----
If just one aggfunc is passed, the name will not be mangled.
"""
if len(aggfuncs) <= 1:
# don't mangle for .agg([lambda x: .])
return aggfuncs
i = 0
mangled_aggfuncs = []
for aggfunc in aggfuncs:
if com.get_callable_name(aggfunc) == "<lambda>":
aggfunc = partial(aggfunc)
aggfunc.__name__ = f"<lambda_{i}>"
i += 1
mangled_aggfuncs.append(aggfunc)
return mangled_aggfuncs
def apply(self, f: F, data: FrameOrSeries, axis: int = 0):
mutated = self.mutated
splitter = self._get_splitter(data, axis=axis)
group_keys = self._get_group_keys()
result_values = None
sdata: FrameOrSeries = splitter._get_sorted_data()
if sdata.ndim == 2 and np.any(
sdata.dtypes.apply(is_extension_array_dtype)):
# calling splitter.fast_apply will raise TypeError via apply_frame_axis0
# if we pass EA instead of ndarray
# TODO: can we have a workaround for EAs backed by ndarray?
pass
elif (com.get_callable_name(f) not in base.plotting_methods
and isinstance(splitter, FrameSplitter) and axis == 0
# fast_apply/libreduction doesn't allow non-numpy backed indexes
and not sdata.index._has_complex_internals):
try:
result_values, mutated = splitter.fast_apply(
f, sdata, group_keys)
except libreduction.InvalidApply as err:
# This Exception is raised if `f` triggers an exception
# but it is preferable to raise the exception in Python.
if "Let this error raise above us" not in str(err):
# TODO: can we infer anything about whether this is
# worth-retrying in pure-python?
raise
else:
# If the fast apply path could be used we can return here.
# Otherwise we need to fall back to the slow implementation.
if len(result_values) == len(group_keys):
return group_keys, result_values, mutated
for key, (i, group) in zip(group_keys, splitter):
object.__setattr__(group, "name", key)
# result_values is None if fast apply path wasn't taken
# or fast apply aborted with an unexpected exception.
# In either case, initialize the result list and perform
# the slow iteration.
if result_values is None:
result_values = []
# If result_values is not None we're in the case that the
# fast apply loop was broken prematurely but we have
# already the result for the first group which we can reuse.
elif i == 0:
continue
# group might be modified
group_axes = group.axes
res = f(group)
if not _is_indexed_like(res, group_axes):
mutated = True
result_values.append(res)
return group_keys, result_values, mutated
def apply(self, f, data, axis=0):
mutated = self.mutated
splitter = self._get_splitter(data, axis=axis)
group_keys = self._get_group_keys()
# oh boy
f_name = com.get_callable_name(f)
if (f_name not in base.plotting_methods
and hasattr(splitter, 'fast_apply') and axis == 0):
try:
values, mutated = splitter.fast_apply(f, group_keys)
return group_keys, values, mutated
except reduction.InvalidApply:
# we detect a mutation of some kind
# so take slow path
pass
except Exception:
# raise this error to the caller
pass
result_values = []
for key, (i, group) in zip(group_keys, splitter):
object.__setattr__(group, 'name', key)
# group might be modified
group_axes = _get_axes(group)
res = f(group)
if not _is_indexed_like(res, group_axes):
mutated = True
result_values.append(res)
return group_keys, result_values, mutated
def apply(self, f, data, axis=0):
mutated = self.mutated
splitter = self._get_splitter(data, axis=axis)
group_keys = self._get_group_keys()
# oh boy
f_name = com.get_callable_name(f)
if (f_name not in base.plotting_methods and
hasattr(splitter, 'fast_apply') and axis == 0):
try:
values, mutated = splitter.fast_apply(f, group_keys)
return group_keys, values, mutated
except reduction.InvalidApply:
# we detect a mutation of some kind
# so take slow path
pass
except Exception:
# raise this error to the caller
pass
result_values = []
for key, (i, group) in zip(group_keys, splitter):
object.__setattr__(group, 'name', key)
# group might be modified
group_axes = _get_axes(group)
res = f(group)
if not _is_indexed_like(res, group_axes):
mutated = True
result_values.append(res)
return group_keys, result_values, mutated
def transform(self) -> FrameOrSeriesUnion:
"""
Transform a DataFrame or Series.
Returns
-------
DataFrame or Series
Result of applying ``func`` along the given axis of the
Series or DataFrame.
Raises
------
ValueError
If the transform function fails or does not transform.
"""
obj = self.obj
func = self.orig_f
axis = self.axis
args = self.args
kwargs = self.kwargs
is_series = obj.ndim == 1
if obj._get_axis_number(axis) == 1:
assert not is_series
return obj.T.transform(func, 0, *args, **kwargs).T
if is_list_like(func) and not is_dict_like(func):
func = cast(List[AggFuncTypeBase], func)
# Convert func equivalent dict
if is_series:
func = {com.get_callable_name(v) or v: v for v in func}
else:
func = {col: func for col in obj}
if is_dict_like(func):
func = cast(AggFuncTypeDict, func)
return self.transform_dict_like(func)
# func is either str or callable
func = cast(AggFuncTypeBase, func)
try:
result = self.transform_str_or_callable(func)
except Exception:
raise ValueError("Transform function failed")
# Functions that transform may return empty Series/DataFrame
# when the dtype is not appropriate
if (
isinstance(result, (ABCSeries, ABCDataFrame))
and result.empty
and not obj.empty
):
raise ValueError("Transform function failed")
if not isinstance(result, (ABCSeries, ABCDataFrame)) or not result.index.equals(
obj.index
):
raise ValueError("Function did not transform")
return result
def transform(obj: FrameOrSeries, func: AggFuncType, axis: Axis, *args,
**kwargs) -> FrameOrSeries:
"""
Transform a DataFrame or Series
Parameters
----------
obj : DataFrame or Series
Object to compute the transform on.
func : string, function, list, or dictionary
Function(s) to compute the transform with.
axis : {0 or 'index', 1 or 'columns'}
Axis along which the function is applied:
* 0 or 'index': apply function to each column.
* 1 or 'columns': apply function to each row.
Returns
-------
DataFrame or Series
Result of applying ``func`` along the given axis of the
Series or DataFrame.
Raises
------
ValueError
If the transform function fails or does not transform.
"""
is_series = obj.ndim == 1
if obj._get_axis_number(axis) == 1:
assert not is_series
return transform(obj.T, func, 0, *args, **kwargs).T
if isinstance(func, list):
if is_series:
func = {com.get_callable_name(v) or v: v for v in func}
else:
func = {col: func for col in obj}
if isinstance(func, dict):
return transform_dict_like(obj, func, *args, **kwargs)
# func is either str or callable
try:
result = transform_str_or_callable(obj, func, *args, **kwargs)
except Exception:
raise ValueError("Transform function failed")
# Functions that transform may return empty Series/DataFrame
# when the dtype is not appropriate
if isinstance(result, (ABCSeries, ABCDataFrame)) and result.empty:
raise ValueError("Transform function failed")
if not isinstance(result,
(ABCSeries, ABCDataFrame)) or not result.index.equals(
obj.index):
raise ValueError("Function did not transform")
return result
def aggregation(
input_df: pd.DataFrame,
group_key: str,
group_values: List[str],
agg_methods: List[Union[str, FunctionType]],
) -> Tuple[pd.DataFrame, List[str]]:
"""
Aggregate values after grouping table rows by a given key.
Args:
input_df:
Input data frame.
group_key:
Used to determine the groups for the groupby.
group_values:
Used to aggregate values for the groupby.
agg_methods:
List of function or function names, e.g. ['mean', 'max', 'min', numpy.mean].
Do not use a lambda function because the name attribute of the lambda function cannot generate a unique string of column names in <lambda>.
Returns:
Tuple of output dataframe and new column names.
"""
new_df = input_df.copy()
new_cols = []
for agg_method in agg_methods:
if _is_lambda_function(agg_method):
raise ValueError('Not supported lambda function.')
elif isinstance(agg_method, str):
pass
elif isinstance(agg_method, FunctionType):
pass
else:
raise ValueError('Supported types are: {} or {}.'
' Got {} instead.'.format(str, Callable,
type(agg_method)))
for agg_method in agg_methods:
for col in group_values:
# only str or FunctionType
if isinstance(agg_method, str):
agg_method_name = agg_method
else:
agg_method_name = get_callable_name(agg_method)
new_col = "agg_{}_{}_by_{}".format(agg_method_name, col, group_key)
df_agg = (input_df[[col] + [group_key]].groupby(group_key)[[
col
]].agg(agg_method))
df_agg.columns = [new_col]
new_cols.append(new_col)
new_df = new_df.merge(df_agg,
how="left",
right_index=True,
left_on=group_key)
return new_df, new_cols
def apply(self, f, data, axis=0):
mutated = self.mutated
splitter = self._get_splitter(data, axis=axis)
group_keys = self._get_group_keys()
result_values = None
# oh boy
f_name = com.get_callable_name(f)
if (
f_name not in base.plotting_methods
and hasattr(splitter, "fast_apply")
and axis == 0
):
try:
result_values, mutated = splitter.fast_apply(f, group_keys)
# If the fast apply path could be used we can return here.
# Otherwise we need to fall back to the slow implementation.
if len(result_values) == len(group_keys):
return group_keys, result_values, mutated
except libreduction.InvalidApply:
# Cannot fast apply on MultiIndex (_has_complex_internals).
# This Exception is also raised if `f` triggers an exception
# but it is preferable to raise the exception in Python.
pass
except TypeError as err:
if "Cannot convert" in str(err):
# via apply_frame_axis0 if we pass a non-ndarray
pass
else:
raise
for key, (i, group) in zip(group_keys, splitter):
object.__setattr__(group, "name", key)
# result_values is None if fast apply path wasn't taken
# or fast apply aborted with an unexpected exception.
# In either case, initialize the result list and perform
# the slow iteration.
if result_values is None:
result_values = []
# If result_values is not None we're in the case that the
# fast apply loop was broken prematurely but we have
# already the result for the first group which we can reuse.
elif i == 0:
continue
# group might be modified
group_axes = _get_axes(group)
res = f(group)
if not _is_indexed_like(res, group_axes):
mutated = True
result_values.append(res)
return group_keys, result_values, mutated
def apply(self, f, data, axis=0):
mutated = self.mutated
splitter = self._get_splitter(data, axis=axis)
group_keys = self._get_group_keys()
result_values = None
# oh boy
f_name = com.get_callable_name(f)
if (f_name not in base.plotting_methods and
hasattr(splitter, 'fast_apply') and axis == 0):
try:
result_values, mutated = splitter.fast_apply(f, group_keys)
# If the fast apply path could be used we can return here.
# Otherwise we need to fall back to the slow implementation.
if len(result_values) == len(group_keys):
return group_keys, result_values, mutated
except reduction.InvalidApply:
# Cannot fast apply on MultiIndex (_has_complex_internals).
# This Exception is also raised if `f` triggers an exception
# but it is preferable to raise the exception in Python.
pass
except Exception:
# raise this error to the caller
pass
for key, (i, group) in zip(group_keys, splitter):
object.__setattr__(group, 'name', key)
# result_values is None if fast apply path wasn't taken
# or fast apply aborted with an unexpected exception.
# In either case, initialize the result list and perform
# the slow iteration.
if result_values is None:
result_values = []
# If result_values is not None we're in the case that the
# fast apply loop was broken prematurely but we have
# already the result for the first group which we can reuse.
elif i == 0:
continue
# group might be modified
group_axes = _get_axes(group)
res = f(group)
if not _is_indexed_like(res, group_axes):
mutated = True
result_values.append(res)
return group_keys, result_values, mutated
def agg_list_like(self) -> DataFrame | Series:
"""
Compute aggregation in the case of a list-like argument.
Returns
-------
Result of aggregation.
"""
from pandas.core.reshape.concat import concat
obj = self.obj
arg = cast(List[AggFuncTypeBase], self.f)
if not isinstance(obj, SelectionMixin):
# i.e. obj is Series or DataFrame
selected_obj = obj
elif obj._selected_obj.ndim == 1:
# For SeriesGroupBy this matches _obj_with_exclusions
selected_obj = obj._selected_obj
else:
selected_obj = obj._obj_with_exclusions
results = []
keys = []
failed_names = []
depr_nuisance_columns_msg = (
"{} did not aggregate successfully. If any error is "
"raised this will raise in a future version of pandas. "
"Drop these columns/ops to avoid this warning.")
# degenerate case
if selected_obj.ndim == 1:
for a in arg:
colg = obj._gotitem(selected_obj.name,
ndim=1,
subset=selected_obj)
try:
new_res = colg.aggregate(a)
except TypeError:
failed_names.append(com.get_callable_name(a) or a)
else:
results.append(new_res)
# make sure we find a good name
name = com.get_callable_name(a) or a
keys.append(name)
# multiples
else:
indices = []
for index, col in enumerate(selected_obj):
colg = obj._gotitem(col,
ndim=1,
subset=selected_obj.iloc[:, index])
try:
# Capture and suppress any warnings emitted by us in the call
# to agg below, but pass through any warnings that were
# generated otherwise.
# This is necessary because of https://bugs.python.org/issue29672
# See GH #43741 for more details
with warnings.catch_warnings(record=True) as record:
new_res = colg.aggregate(arg)
if len(record) > 0:
match = re.compile(
depr_nuisance_columns_msg.format(".*"))
for warning in record:
if re.match(match, str(warning.message)):
failed_names.append(col)
else:
warnings.warn_explicit(
message=warning.message,
category=warning.category,
filename=warning.filename,
lineno=warning.lineno,
)
except (TypeError, DataError):
failed_names.append(col)
except ValueError as err:
# cannot aggregate
if "Must produce aggregated value" in str(err):
# raised directly in _aggregate_named
failed_names.append(col)
elif "no results" in str(err):
# reached in test_frame_apply.test_nuiscance_columns
# where the colg.aggregate(arg) ends up going through
# the selected_obj.ndim == 1 branch above with arg == ["sum"]
# on a datetime64[ns] column
failed_names.append(col)
else:
raise
else:
results.append(new_res)
indices.append(index)
keys = selected_obj.columns.take(indices)
# if we are empty
if not len(results):
raise ValueError("no results")
if len(failed_names) > 0:
warnings.warn(
depr_nuisance_columns_msg.format(failed_names),
FutureWarning,
stacklevel=find_stack_level(),
)
try:
concatenated = concat(results, keys=keys, axis=1, sort=False)
except TypeError as err:
# we are concatting non-NDFrame objects,
# e.g. a list of scalars
from pandas import Series
result = Series(results, index=keys, name=obj.name)
if is_nested_object(result):
raise ValueError(
"cannot combine transform and aggregation operations"
) from err
return result
else:
# Concat uses the first index to determine the final indexing order.
# The union of a shorter first index with the other indices causes
# the index sorting to be different from the order of the aggregating
# functions. Reindex if this is the case.
index_size = concatenated.index.size
full_ordered_index = next(result.index for result in results
if result.index.size == index_size)
return concatenated.reindex(full_ordered_index, copy=False)
def relabel_result(
result: DataFrame | Series,
func: dict[str, list[Callable | str]],
columns: Iterable[Hashable],
order: Iterable[int],
) -> dict[Hashable, Series]:
"""
Internal function to reorder result if relabelling is True for
dataframe.agg, and return the reordered result in dict.
Parameters:
----------
result: Result from aggregation
func: Dict of (column name, funcs)
columns: New columns name for relabelling
order: New order for relabelling
Examples:
---------
>>> result = DataFrame({"A": [np.nan, 2, np.nan],
... "C": [6, np.nan, np.nan], "B": [np.nan, 4, 2.5]}) # doctest: +SKIP
>>> funcs = {"A": ["max"], "C": ["max"], "B": ["mean", "min"]}
>>> columns = ("foo", "aab", "bar", "dat")
>>> order = [0, 1, 2, 3]
>>> _relabel_result(result, func, columns, order) # doctest: +SKIP
dict(A=Series([2.0, NaN, NaN, NaN], index=["foo", "aab", "bar", "dat"]),
C=Series([NaN, 6.0, NaN, NaN], index=["foo", "aab", "bar", "dat"]),
B=Series([NaN, NaN, 2.5, 4.0], index=["foo", "aab", "bar", "dat"]))
"""
from pandas.core.indexes.base import Index
reordered_indexes = [
pair[0] for pair in sorted(zip(columns, order), key=lambda t: t[1])
]
reordered_result_in_dict: dict[Hashable, Series] = {}
idx = 0
reorder_mask = not isinstance(result,
ABCSeries) and len(result.columns) > 1
for col, fun in func.items():
s = result[col].dropna()
# In the `_aggregate`, the callable names are obtained and used in `result`, and
# these names are ordered alphabetically. e.g.
# C2 C1
# <lambda> 1 NaN
# amax NaN 4.0
# max NaN 4.0
# sum 18.0 6.0
# Therefore, the order of functions for each column could be shuffled
# accordingly so need to get the callable name if it is not parsed names, and
# reorder the aggregated result for each column.
# e.g. if df.agg(c1=("C2", sum), c2=("C2", lambda x: min(x))), correct order is
# [sum, <lambda>], but in `result`, it will be [<lambda>, sum], and we need to
# reorder so that aggregated values map to their functions regarding the order.
# However there is only one column being used for aggregation, not need to
# reorder since the index is not sorted, and keep as is in `funcs`, e.g.
# A
# min 1.0
# mean 1.5
# mean 1.5
if reorder_mask:
fun = [
com.get_callable_name(f) if not isinstance(f, str) else f
for f in fun
]
col_idx_order = Index(s.index).get_indexer(fun)
s = s[col_idx_order]
# assign the new user-provided "named aggregation" as index names, and reindex
# it based on the whole user-provided names.
s.index = reordered_indexes[idx:idx + len(fun)]
reordered_result_in_dict[col] = s.reindex(columns, copy=False)
idx = idx + len(fun)
return reordered_result_in_dict
def _aggregate_multiple_funcs(self, arg, _axis):
from pandas.core.reshape.concat import concat
if _axis != 0:
raise NotImplementedError("axis other than 0 is not supported")
if self._selected_obj.ndim == 1:
obj = self._selected_obj
else:
obj = self._obj_with_exclusions
results = []
keys = []
# degenerate case
if obj.ndim == 1:
for a in arg:
colg = self._gotitem(obj.name, ndim=1, subset=obj)
try:
new_res = colg.aggregate(a)
except TypeError:
pass
else:
results.append(new_res)
# make sure we find a good name
name = com.get_callable_name(a) or a
keys.append(name)
# multiples
else:
for index, col in enumerate(obj):
colg = self._gotitem(col, ndim=1, subset=obj.iloc[:, index])
try:
new_res = colg.aggregate(arg)
except (TypeError, DataError):
pass
except ValueError as err:
# cannot aggregate
if "Must produce aggregated value" in str(err):
# raised directly in _aggregate_named
pass
elif "no results" in str(err):
# raised direcly in _aggregate_multiple_funcs
pass
else:
raise
else:
results.append(new_res)
keys.append(col)
# if we are empty
if not len(results):
raise ValueError("no results")
try:
return concat(results, keys=keys, axis=1, sort=False)
except TypeError as err:
# we are concatting non-NDFrame objects,
# e.g. a list of scalars
from pandas import Series
result = Series(results, index=keys, name=self.name)
if is_nested_object(result):
raise ValueError(
"cannot combine transform and aggregation operations"
) from err
return result
def _aggregate_multiple_funcs(self, arg, _level, _axis):
from pandas.core.reshape.concat import concat
if _axis != 0:
raise NotImplementedError("axis other than 0 is not supported")
if self._selected_obj.ndim == 1:
obj = self._selected_obj
else:
obj = self._obj_with_exclusions
results = []
keys = []
# degenerate case
if obj.ndim == 1:
for a in arg:
try:
colg = self._gotitem(obj.name, ndim=1, subset=obj)
results.append(colg.aggregate(a))
# make sure we find a good name
name = com.get_callable_name(a) or a
keys.append(name)
except (TypeError, DataError):
pass
except SpecificationError:
raise
# multiples
else:
for index, col in enumerate(obj):
try:
colg = self._gotitem(col,
ndim=1,
subset=obj.iloc[:, index])
results.append(colg.aggregate(arg))
keys.append(col)
except (TypeError, DataError):
pass
except ValueError:
# cannot aggregate
continue
except SpecificationError:
raise
# if we are empty
if not len(results):
raise ValueError("no results")
try:
return concat(results, keys=keys, axis=1, sort=False)
except TypeError:
# we are concatting non-NDFrame objects,
# e.g. a list of scalars
from pandas.core.dtypes.cast import is_nested_object
from pandas import Series
result = Series(results, index=keys, name=self.name)
if is_nested_object(result):
raise ValueError("cannot combine transform and "
"aggregation operations")
return result
def agg_list_like(self) -> FrameOrSeriesUnion:
"""
Compute aggregation in the case of a list-like argument.
Returns
-------
Result of aggregation.
"""
from pandas.core.reshape.concat import concat
obj = self.obj
arg = cast(List[AggFuncTypeBase], self.f)
if not isinstance(obj, SelectionMixin):
# i.e. obj is Series or DataFrame
selected_obj = obj
elif obj._selected_obj.ndim == 1:
# For SeriesGroupBy this matches _obj_with_exclusions
selected_obj = obj._selected_obj
else:
selected_obj = obj._obj_with_exclusions
results = []
keys = []
# degenerate case
if selected_obj.ndim == 1:
for a in arg:
colg = obj._gotitem(selected_obj.name, ndim=1, subset=selected_obj)
try:
new_res = colg.aggregate(a)
except TypeError:
pass
else:
results.append(new_res)
# make sure we find a good name
name = com.get_callable_name(a) or a
keys.append(name)
# multiples
else:
indices = []
for index, col in enumerate(selected_obj):
colg = obj._gotitem(col, ndim=1, subset=selected_obj.iloc[:, index])
try:
new_res = colg.aggregate(arg)
except (TypeError, DataError):
pass
except ValueError as err:
# cannot aggregate
if "Must produce aggregated value" in str(err):
# raised directly in _aggregate_named
pass
elif "no results" in str(err):
# reached in test_frame_apply.test_nuiscance_columns
# where the colg.aggregate(arg) ends up going through
# the selected_obj.ndim == 1 branch above with arg == ["sum"]
# on a datetime64[ns] column
pass
else:
raise
else:
results.append(new_res)
indices.append(index)
keys = selected_obj.columns.take(indices)
# if we are empty
if not len(results):
raise ValueError("no results")
try:
concatenated = concat(results, keys=keys, axis=1, sort=False)
except TypeError as err:
# we are concatting non-NDFrame objects,
# e.g. a list of scalars
from pandas import Series
result = Series(results, index=keys, name=obj.name)
if is_nested_object(result):
raise ValueError(
"cannot combine transform and aggregation operations"
) from err
return result
else:
# Concat uses the first index to determine the final indexing order.
# The union of a shorter first index with the other indices causes
# the index sorting to be different from the order of the aggregating
# functions. Reindex if this is the case.
index_size = concatenated.index.size
full_ordered_index = next(
result.index for result in results if result.index.size == index_size
)
return concatenated.reindex(full_ordered_index, copy=False)
def transform(obj: FrameOrSeries, func: AggFuncType, axis: Axis, *args,
**kwargs) -> FrameOrSeries:
"""
Transform a DataFrame or Series
Parameters
----------
obj : DataFrame or Series
Object to compute the transform on.
func : string, function, list, or dictionary
Function(s) to compute the transform with.
axis : {0 or 'index', 1 or 'columns'}
Axis along which the function is applied:
* 0 or 'index': apply function to each column.
* 1 or 'columns': apply function to each row.
Returns
-------
DataFrame or Series
Result of applying ``func`` along the given axis of the
Series or DataFrame.
Raises
------
ValueError
If the transform function fails or does not transform.
"""
from pandas.core.reshape.concat import concat
is_series = obj.ndim == 1
if obj._get_axis_number(axis) == 1:
assert not is_series
return transform(obj.T, func, 0, *args, **kwargs).T
if isinstance(func, list):
if is_series:
func = {com.get_callable_name(v) or v: v for v in func}
else:
func = {col: func for col in obj}
if isinstance(func, dict):
if not is_series:
cols = sorted(set(func.keys()) - set(obj.columns))
if len(cols) > 0:
raise SpecificationError(f"Column(s) {cols} do not exist")
if any(isinstance(v, dict) for v in func.values()):
# GH 15931 - deprecation of renaming keys
raise SpecificationError("nested renamer is not supported")
results = {}
for name, how in func.items():
colg = obj._gotitem(name, ndim=1)
try:
results[name] = transform(colg, how, 0, *args, **kwargs)
except Exception as e:
if str(e) == "Function did not transform":
raise e
# combine results
if len(results) == 0:
raise ValueError("Transform function failed")
return concat(results, axis=1)
# func is either str or callable
try:
if isinstance(func, str):
result = obj._try_aggregate_string_function(func, *args, **kwargs)
else:
f = obj._get_cython_func(func)
if f and not args and not kwargs:
result = getattr(obj, f)()
else:
try:
result = obj.apply(func, args=args, **kwargs)
except Exception:
result = func(obj, *args, **kwargs)
except Exception:
raise ValueError("Transform function failed")
# Functions that transform may return empty Series/DataFrame
# when the dtype is not appropriate
if isinstance(result, (ABCSeries, ABCDataFrame)) and result.empty:
raise ValueError("Transform function failed")
if not isinstance(result,
(ABCSeries, ABCDataFrame)) or not result.index.equals(
obj.index):
raise ValueError("Function did not transform")
return result
def agg_list_like(self) -> FrameOrSeriesUnion:
"""
Compute aggregation in the case of a list-like argument.
Returns
-------
Result of aggregation.
"""
from pandas.core.reshape.concat import concat
obj = self.obj
arg = cast(List[AggFuncTypeBase], self.f)
if not isinstance(obj, SelectionMixin):
# i.e. obj is Series or DataFrame
selected_obj = obj
elif obj._selected_obj.ndim == 1:
selected_obj = obj._selected_obj
else:
selected_obj = obj._obj_with_exclusions
results = []
keys = []
# degenerate case
if selected_obj.ndim == 1:
for a in arg:
colg = obj._gotitem(selected_obj.name,
ndim=1,
subset=selected_obj)
try:
new_res = colg.aggregate(a)
except TypeError:
pass
else:
results.append(new_res)
# make sure we find a good name
name = com.get_callable_name(a) or a
keys.append(name)
# multiples
else:
for index, col in enumerate(selected_obj):
colg = obj._gotitem(col,
ndim=1,
subset=selected_obj.iloc[:, index])
try:
new_res = colg.aggregate(arg)
except (TypeError, DataError):
pass
except ValueError as err:
# cannot aggregate
if "Must produce aggregated value" in str(err):
# raised directly in _aggregate_named
pass
elif "no results" in str(err):
# raised directly in _aggregate_multiple_funcs
pass
else:
raise
else:
results.append(new_res)
keys.append(col)
# if we are empty
if not len(results):
raise ValueError("no results")
try:
return concat(results, keys=keys, axis=1, sort=False)
except TypeError as err:
# we are concatting non-NDFrame objects,
# e.g. a list of scalars
from pandas import Series
result = Series(results, index=keys, name=obj.name)
if is_nested_object(result):
raise ValueError(
"cannot combine transform and aggregation operations"
) from err
return result
def _aggregate_multiple_funcs(self, arg, _level, _axis):
from pandas.core.reshape.concat import concat
if _axis != 0:
raise NotImplementedError("axis other than 0 is not supported")
if self._selected_obj.ndim == 1:
obj = self._selected_obj
else:
obj = self._obj_with_exclusions
results = []
keys = []
# degenerate case
if obj.ndim == 1:
for a in arg:
try:
colg = self._gotitem(obj.name, ndim=1, subset=obj)
results.append(colg.aggregate(a))
# make sure we find a good name
name = com.get_callable_name(a) or a
keys.append(name)
except (TypeError, DataError):
pass
except SpecificationError:
raise
# multiples
else:
for index, col in enumerate(obj):
try:
colg = self._gotitem(col, ndim=1,
subset=obj.iloc[:, index])
results.append(colg.aggregate(arg))
keys.append(col)
except (TypeError, DataError):
pass
except ValueError:
# cannot aggregate
continue
except SpecificationError:
raise
# if we are empty
if not len(results):
raise ValueError("no results")
try:
return concat(results, keys=keys, axis=1, sort=False)
except TypeError:
# we are concatting non-NDFrame objects,
# e.g. a list of scalars
from pandas.core.dtypes.cast import is_nested_object
from pandas import Series
result = Series(results, index=keys, name=self.name)
if is_nested_object(result):
raise ValueError("cannot combine transform and "
"aggregation operations")
return result
def apply(self, f: F, data: FrameOrSeries, axis: int = 0):
mutated = self.mutated
splitter = self._get_splitter(data, axis=axis)
group_keys = self._get_group_keys()
result_values = None
if data.ndim == 2 and np.any(
data.dtypes.apply(is_extension_array_dtype)):
# calling splitter.fast_apply will raise TypeError via apply_frame_axis0
# if we pass EA instead of ndarray
# TODO: can we have a workaround for EAs backed by ndarray?
pass
elif isinstance(data._mgr, ArrayManager):
# TODO(ArrayManager) don't use fast_apply / libreduction.apply_frame_axis0
# for now -> relies on BlockManager internals
pass
elif (com.get_callable_name(f) not in base.plotting_methods
and isinstance(splitter, FrameSplitter) and axis == 0
# fast_apply/libreduction doesn't allow non-numpy backed indexes
and not data.index._has_complex_internals):
try:
sdata = splitter.sorted_data
result_values, mutated = splitter.fast_apply(
f, sdata, group_keys)
except IndexError:
# This is a rare case in which re-running in python-space may
# make a difference, see test_apply_mutate.test_mutate_groups
pass
else:
# If the fast apply path could be used we can return here.
# Otherwise we need to fall back to the slow implementation.
if len(result_values) == len(group_keys):
return group_keys, result_values, mutated
for key, (i, group) in zip(group_keys, splitter):
object.__setattr__(group, "name", key)
# result_values is None if fast apply path wasn't taken
# or fast apply aborted with an unexpected exception.
# In either case, initialize the result list and perform
# the slow iteration.
if result_values is None:
result_values = []
# If result_values is not None we're in the case that the
# fast apply loop was broken prematurely but we have
# already the result for the first group which we can reuse.
elif i == 0:
continue
# group might be modified
group_axes = group.axes
res = f(group)
if not _is_indexed_like(res, group_axes, axis):
mutated = True
result_values.append(res)
return group_keys, result_values, mutated
def transform(self) -> DataFrame | Series:
"""
Transform a DataFrame or Series.
Returns
-------
DataFrame or Series
Result of applying ``func`` along the given axis of the
Series or DataFrame.
Raises
------
ValueError
If the transform function fails or does not transform.
"""
obj = self.obj
func = self.orig_f
axis = self.axis
args = self.args
kwargs = self.kwargs
is_series = obj.ndim == 1
if obj._get_axis_number(axis) == 1:
assert not is_series
return obj.T.transform(func, 0, *args, **kwargs).T
if is_list_like(func) and not is_dict_like(func):
func = cast(List[AggFuncTypeBase], func)
# Convert func equivalent dict
if is_series:
func = {com.get_callable_name(v) or v: v for v in func}
else:
func = {col: func for col in obj}
if is_dict_like(func):
func = cast(AggFuncTypeDict, func)
return self.transform_dict_like(func)
# func is either str or callable
func = cast(AggFuncTypeBase, func)
try:
result = self.transform_str_or_callable(func)
except TypeError:
raise
except Exception as err:
raise ValueError("Transform function failed") from err
# Functions that transform may return empty Series/DataFrame
# when the dtype is not appropriate
if (isinstance(result, (ABCSeries, ABCDataFrame)) and result.empty
and not obj.empty):
raise ValueError("Transform function failed")
# error: Argument 1 to "__get__" of "AxisProperty" has incompatible type
# "Union[Series, DataFrame, GroupBy[Any], SeriesGroupBy,
# DataFrameGroupBy, BaseWindow, Resampler]"; expected "Union[DataFrame,
# Series]"
if not isinstance(
result, (ABCSeries, ABCDataFrame)) or not result.index.equals(
obj.index # type:ignore[arg-type]
):
raise ValueError("Function did not transform")
return result
