def reduce(self,
reduce_func_nb,
*args,
idx_arr=None,
to_array=False,
to_idx=False,
idx_labeled=True,
default_val=np.nan,
group_by=None,
**kwargs):
"""Reduce mapped array by column.
If `to_array` is False and `to_idx` is False, see `vectorbt.records.nb.reduce_mapped_nb`.
If `to_array` is False and `to_idx` is True, see `vectorbt.records.nb.reduce_mapped_to_idx_nb`.
If `to_array` is True and `to_idx` is False, see `vectorbt.records.nb.reduce_mapped_to_array_nb`.
If `to_array` is True and `to_idx` is True, see `vectorbt.records.nb.reduce_mapped_to_idx_array_nb`.
If `to_idx` is True, must pass `idx_arr`. Set `idx_labeled` to False to return raw positions instead
of labels. Use `default_val` to set the default value. Set `group_by` to False to disable grouping.
`**kwargs` will be passed to `vectorbt.base.array_wrapper.ArrayWrapper.wrap_reduced`."""
# Perform checks
checks.assert_numba_func(reduce_func_nb)
if idx_arr is None:
if self.idx_arr is None:
if to_idx:
raise ValueError("Must pass idx_arr")
idx_arr = self.idx_arr
# Perform main computation
col_map = self.col_mapper.get_col_map(group_by=group_by)
if not to_array:
if not to_idx:
out = nb.reduce_mapped_nb(self.values, col_map, default_val,
reduce_func_nb, *args)
else:
out = nb.reduce_mapped_to_idx_nb(self.values, col_map, idx_arr,
default_val, reduce_func_nb,
*args)
else:
if not to_idx:
out = nb.reduce_mapped_to_array_nb(self.values, col_map,
default_val, reduce_func_nb,
*args)
else:
out = nb.reduce_mapped_to_idx_array_nb(self.values, col_map,
idx_arr, default_val,
reduce_func_nb, *args)
# Perform post-processing
if to_idx:
nan_mask = np.isnan(out)
if idx_labeled:
out = out.astype(np.object)
out[~nan_mask] = self.wrapper.index[out[~nan_mask].astype(
np.int_)]
else:
out[nan_mask] = -1
out = out.astype(np.int_)
return self.wrapper.wrap_reduced(out, group_by=group_by, **kwargs)
def reduce(self,
reduce_func_nb: tp.ReduceFunc,
*args,
idx_arr: tp.Optional[tp.Array1d] = None,
returns_array: bool = False,
returns_idx: bool = False,
to_index: bool = True,
fill_value: tp.Scalar = np.nan,
group_by: tp.GroupByLike = None,
wrap_kwargs: tp.KwargsLike = None) -> tp.MaybeSeriesFrame:
"""Reduce mapped array by column/group.
If `returns_array` is False and `returns_idx` is False, see `vectorbt.records.nb.reduce_mapped_nb`.
If `returns_array` is False and `returns_idx` is True, see `vectorbt.records.nb.reduce_mapped_to_idx_nb`.
If `returns_array` is True and `returns_idx` is False, see `vectorbt.records.nb.reduce_mapped_to_array_nb`.
If `returns_array` is True and `returns_idx` is True, see `vectorbt.records.nb.reduce_mapped_to_idx_array_nb`.
If `returns_idx` is True, must pass `idx_arr`. Set `to_index` to False to return raw positions instead
of labels. Use `fill_value` to set the default value. Set `group_by` to False to disable grouping.
"""
# Perform checks
checks.assert_numba_func(reduce_func_nb)
if idx_arr is None:
if self.idx_arr is None:
if returns_idx:
raise ValueError("Must pass idx_arr")
idx_arr = self.idx_arr
# Perform main computation
col_map = self.col_mapper.get_col_map(group_by=group_by)
if not returns_array:
if not returns_idx:
out = nb.reduce_mapped_nb(self.values, col_map, fill_value,
reduce_func_nb, *args)
else:
out = nb.reduce_mapped_to_idx_nb(self.values, col_map, idx_arr,
fill_value, reduce_func_nb,
*args)
else:
if not returns_idx:
out = nb.reduce_mapped_to_array_nb(self.values, col_map,
fill_value, reduce_func_nb,
*args)
else:
out = nb.reduce_mapped_to_idx_array_nb(self.values, col_map,
idx_arr, fill_value,
reduce_func_nb, *args)
# Perform post-processing
wrap_kwargs = merge_dicts(
dict(name_or_index='reduce' if not returns_array else None,
to_index=returns_idx and to_index,
fillna=-1 if returns_idx else None,
dtype=np.int_ if returns_idx else None), wrap_kwargs)
return self.wrapper.wrap_reduced(out, group_by=group_by, **wrap_kwargs)
def reduce(self,
reduce_func_nb,
*args,
default_val=np.nan,
cast=None,
**kwargs):
"""Reduce mapped array by column to a scalar value.
See `vectorbt.records.nb.reduce_mapped_nb`.
`**kwargs` will be passed to `vectorbt.base.array_wrapper.ArrayWrapper.wrap_reduced`."""
checks.assert_numba_func(reduce_func_nb)
result = nb.reduce_mapped_nb(self.mapped_arr, self.col_arr,
len(self.wrapper.columns), default_val,
reduce_func_nb, *args)
if cast is not None:
result = result.astype(cast)
return self.wrapper.wrap_reduced(result, **kwargs)
def reduce(self,
reduce_func_nb: tp.ReduceFunc,
*args,
idx_arr: tp.Optional[tp.Array1d] = None,
to_array: bool = False,
to_idx: bool = False,
idx_labeled: bool = True,
default_val: float = np.nan,
group_by: tp.GroupByLike = None,
wrap_kwargs: tp.KwargsLike = None) -> tp.MaybeSeriesFrame:
"""Reduce mapped array by column.
If `to_array` is False and `to_idx` is False, see `vectorbt.records.nb.reduce_mapped_nb`.
If `to_array` is False and `to_idx` is True, see `vectorbt.records.nb.reduce_mapped_to_idx_nb`.
If `to_array` is True and `to_idx` is False, see `vectorbt.records.nb.reduce_mapped_to_array_nb`.
If `to_array` is True and `to_idx` is True, see `vectorbt.records.nb.reduce_mapped_to_idx_array_nb`.
If `to_idx` is True, must pass `idx_arr`. Set `idx_labeled` to False to return raw positions instead
of labels. Use `default_val` to set the default value. Set `group_by` to False to disable grouping.
"""
# Perform checks
checks.assert_numba_func(reduce_func_nb)
if idx_arr is None:
if self.idx_arr is None:
if to_idx:
raise ValueError("Must pass idx_arr")
idx_arr = self.idx_arr
# Perform main computation
col_map = self.col_mapper.get_col_map(group_by=group_by)
if not to_array:
if not to_idx:
out = nb.reduce_mapped_nb(self.values, col_map, default_val,
reduce_func_nb, *args)
else:
out = nb.reduce_mapped_to_idx_nb(self.values, col_map, idx_arr,
default_val, reduce_func_nb,
*args)
else:
if not to_idx:
out = nb.reduce_mapped_to_array_nb(self.values, col_map,
default_val, reduce_func_nb,
*args)
else:
out = nb.reduce_mapped_to_idx_array_nb(self.values, col_map,
idx_arr, default_val,
reduce_func_nb, *args)
# Perform post-processing
if to_idx:
nan_mask = np.isnan(out)
if idx_labeled:
out = out.astype(object)
out[~nan_mask] = self.wrapper.index[out[~nan_mask].astype(
np.int_)]
else:
out[nan_mask] = -1
out = out.astype(np.int_)
wrap_kwargs = merge_dicts(
dict(name_or_index='reduce' if not to_array else None),
wrap_kwargs)
return self.wrapper.wrap_reduced(out, group_by=group_by, **wrap_kwargs)
def reduce(self, reduce_func_nb, *args, idx_arr=None, to_array=False, n_rows=None, to_idx=False,
idx_labeled=True, default_val=np.nan, cast=None, group_by=None, **kwargs):
"""Reduce mapped array by column.
If `to_array` is `False` and `to_idx` is `False`, see `vectorbt.records.nb.reduce_mapped_nb`.
If `to_array` is `False` and `to_idx` is `True`, see `vectorbt.records.nb.reduce_mapped_to_idx_nb`.
If `to_array` is `True` and `to_idx` is `False`, see `vectorbt.records.nb.reduce_mapped_to_array_nb`.
If `to_array` is `True` and `to_idx` is `True`, see `vectorbt.records.nb.reduce_mapped_to_idx_array_nb`.
If `to_array` is `True`, must pass `n_rows` indicating the number of elements in the array.
If `to_idx` is `True`, must pass `idx_arr`. Set `idx_labeled` to `False` to return raw positions
instead of labels. Use `default_val` to set the default value and `cast` to perform casting
on the resulting pandas object. Set `group_by` to `False` to disable grouping.
`**kwargs` will be passed to `vectorbt.base.array_wrapper.ArrayWrapper.wrap_reduced`."""
# Perform checks
checks.assert_numba_func(reduce_func_nb)
if idx_arr is None:
if self.idx_arr is None:
if to_idx:
raise ValueError("Must pass idx_arr")
idx_arr = self.idx_arr
# Perform main computation
group_arr, columns = self.wrapper.grouper.get_groups_and_columns(group_by=group_by)
if group_arr is not None:
col_arr = group_arr[self.col_arr]
else:
col_arr = self.col_arr
if not to_array:
if not to_idx:
result = nb.reduce_mapped_nb(
self.mapped_arr,
col_arr,
len(columns),
default_val,
reduce_func_nb,
*args
)
else:
result = nb.reduce_mapped_to_idx_nb(
self.mapped_arr,
col_arr,
idx_arr,
len(columns),
default_val,
reduce_func_nb,
*args
)
else:
checks.assert_not_none(n_rows)
if not to_idx:
result = nb.reduce_mapped_to_array_nb(
self.mapped_arr,
col_arr,
len(columns),
n_rows,
default_val,
reduce_func_nb,
*args
)
else:
result = nb.reduce_mapped_to_idx_array_nb(
self.mapped_arr,
col_arr,
idx_arr,
len(columns),
n_rows,
default_val,
reduce_func_nb,
*args
)
# Perform post-processing
if to_idx:
if idx_labeled:
result_shape = result.shape
result = result.flatten()
mask = np.isnan(result)
if mask.any():
# Contains NaNs
result[mask] = 0
result = result.astype(int)
result = self.wrapper.index[result].to_numpy()
result = result.astype(np.object)
result[mask] = np.nan
else:
result = self.wrapper.index[result.astype(int)].to_numpy()
result = np.reshape(result, result_shape)
else:
mask = np.isnan(result)
result[mask] = -1
result = result.astype(int)
result = self.wrapper.wrap_reduced(result, group_by=group_by, **kwargs)
if cast is not None:
result = result.astype(cast)
return result