torcharrow.column

torcharrow.column(data: Optional[Union[Iterable, DType]] = None, dtype: Optional[DType] = None, device: str = '')

Creates a TorchArrow Column. Allocates memory on device or default device.

Parameters:

• data (array-like or Iterable) – Defines the contents of the column.

• dtype (dtype, default None) – Data type to force. If None the type will be automatically inferred where possible.

• device (Device, default "") – Device selects which runtime to use from scope. TorchArrow supports multiple runtimes (CPU and GPU). If not supplied, uses the Velox vectorized runtime. Valid values are “cpu” (Velox), “gpu” (coming soon).

Examples

Creating a column using auto-inferred type:

>>> import torcharrow as ta
>>> s = ta.column([1,2,None,4])
>>> s
0  1
1  2
2  None
3  4
dtype: Int64(nullable=True), length: 4, null_count: 1

Create a column with arbitrarily data types, here a non-nullable column of a list of non-nullable strings of arbitrary length:

>>> sf = ta.column([ ["hello", "world"], ["how", "are", "you"] ], dtype =dt.List(dt.string))
>>> sf.dtype
List(item_dtype=String(nullable=False), nullable=False, fixed_size=-1)

Create a column of average climate data, one map per continent, with city as key and yearly average min and max temperature:

>>> mf = ta.column([
>>>     {'helsinki': [-1.3, 21.5], 'moscow': [-4.0,24.3]},
>>>     {'algiers':[11.2, 25.2], 'kinshasa':[22.2,26.8]}
>>>     ])
>>> mf
0  {'helsinki': [-1.3, 21.5], 'moscow': [-4.0, 24.3]}
1  {'algiers': [11.2, 25.2], 'kinshasa': [22.2, 26.8]}
dtype: Map(string, List(float64)), length: 2, null_count: 0