replace#

pylibcudf.replace.ReplacePolicy#

See also replace_policy.

Enum members

PRECEDING
FOLLOWING

pylibcudf.replace.clamp( Column source_column, Scalar lo, Scalar hi, Scalar lo_replace=None, Scalar hi_replace=None, stream=None, DeviceMemoryResource mr=None, ) → Column#

Clamp the values in source_column to the range [lo, hi].

For details, see clamp().

Parameters:

source_columnColumn: The column to clamp.
loScalar: The lower bound of the clamp range.
hiScalar: The upper bound of the clamp range.
lo_replaceScalar, optional: The value to use for elements that are less than lo. If not specified, the value of lo is used.
hi_replaceScalar, optional: The value to use for elements that are greater than hi. If not specified, the value of hi is used.
streamStream | None: CUDA stream on which to perform the operation.
mrDeviceMemoryResource | None: Device memory resource used to allocate the returned column’s device memory.

Returns:

Column: A copy of source_column with values clamped to the range [lo, hi].

pylibcudf.replace.find_and_replace_all( Column source_column, Column values_to_replace, Column replacement_values, stream=None, DeviceMemoryResource mr=None, ) → Column#

Replace all occurrences of values_to_replace with replacement_values.

For details, see find_and_replace_all().

Parameters:

source_columnColumn: The column in which to replace values.
values_to_replaceColumn: The column containing values to replace.
replacement_valuesColumn: The column containing replacement values.
streamStream | None: CUDA stream on which to perform the operation.
mrDeviceMemoryResource | None: Device memory resource used to allocate the returned column’s device memory.

Returns:

Column: A copy of source_column with all occurrences of values_to_replace replaced by replacement_values.

pylibcudf.replace.normalize_nans_and_zeros( Column source_column, bool inplace=False, stream=None, DeviceMemoryResource mr=None, ) → Column#

Normalize NaNs and zeros in source_column.

For details, see normalize_nans_and_zeros().

Parameters:

source_columnColumn: The column to normalize.
inplacebool, optional: If True, normalize source_column in place. If False, return a new column with the normalized values.
streamStream | None: CUDA stream on which to perform the operation.
mrDeviceMemoryResource | None: Device memory resource used to allocate the returned column’s device memory.

Returns:

Column: A copy of source_column with NaNs and zeros normalized.

pylibcudf.replace.replace_nulls(signatures, args, kwargs, defaults, _fused_sigindex={})#

Replace nulls in source_column.

The values used to replace nulls depends on the type of replacement:

If replacement is a Column, the corresponding value from replacement is used.
If replacement is a Scalar, the same value is used for all nulls.
If replacement is a replace_policy, the policy is used to determine the replacement value:
- PRECEDING: The first non-null value that precedes the null is used.
- FOLLOWING: The first non-null value that follows the null is used.

For more details, see replace_nulls().

Parameters:

source_columnColumn: The column in which to replace nulls.
replacement_columnUnion[Column, Scalar, replace_policy]: If a Column, the values to use as replacements. If a Scalar, the value to use as a replacement. If a replace_policy, the policy to use to determine the replacement value.
streamStream | None: CUDA stream on which to perform the operation.
mrDeviceMemoryResource | None: Device memory resource used to allocate the returned column’s device memory.

Returns:

Column: A copy of source_column with nulls replaced by values from replacement_column.