pandas.Index.take#

Index.take(indices, axis=0, allow_fill=<no_default>, fill_value=<no_default>, **kwargs)[source]#

Return a new Index of the values selected by the indices.

For internal compatibility with numpy arrays.

Parameters:
indicesarray-like

Indices to be taken.

axis{0 or ‘index’}, optional

The axis over which to select values, always 0 or ‘index’.

allow_fillbool, optional

How to handle negative values in indices.

  • False: negative values in indices indicate positional indices from the right, matching numpy.take().

  • True: negative values in indices indicate missing values. -1 entries are set to fill_value (using the default NA value of the caller’s dtype if not supplied). Any other negative values raise a ValueError.

  • Not supplied: defaults to allow_fill=False unless fill_value is explicitly provided, in which case fill semantics apply (allow_fill=True).

fill_valuescalar, optional

If fill semantics apply (see allow_fill), indices specified by -1 are filled with fill_value. Not specifying fill_value or passing fill_value=None will fill using the default NA value of the caller’s dtype. If the Index’s dtype cannot hold fill_value, the result is promoted to a common dtype (e.g. an integer Index + np.nan fill is cast to float, while an integer Index + a string fill is cast to object).

**kwargs

Required for compatibility with numpy.

Returns:
Index

An index formed of elements at the given indices. Will be the same type as self, except for RangeIndex.

See also

numpy.ndarray.take

Return an array formed from the elements of a at the given indices.

Examples

>>> idx = pd.Index(["a", "b", "c"])
>>> idx.take([2, 2, 1, 2])
Index(['c', 'c', 'b', 'c'], dtype='str')
On this page