torcharrow.functional.bucketize

torcharrow.functional.bucketize(value_col: NumericalColumn, borders: Union[ListColumn, List[Union[int, float]]]) → NumericalColumn

Apply bucketization for input feature. This is a common operation in recommendation domain to convert dense features into sparse features.

Parameters:

• value_col (Numeric column that defines dense feature) –

• borders (Border values for the discretized sparse features) –

Examples

>>> import torcharrow as ta
>>> from torcharrow import functional
>>> a = ta.column([1, 2, 3, 5, 8, 10, 11])
>>> functional.bucketize(a, [2, 5, 10])
0  0
1  0
2  1
3  1
4  2
5  2
6  3
dtype: Int32(nullable=True), length: 7, null_count: 0