BUG: allow complex type in convert_dtypes

[yasirroni]

Feature Type

• Adding new functionality to pandas

• Changing existing functionality in pandas

• Removing existing functionality in pandas

Problem Description

convert_dtypes should be able to do this type of coercion from string complex looking

print(pd.DataFrame(['1.0+5j', '1.5-3j']).astype(complex).convert_dtypes())

Related to closed #4895

Feature Description

Alternative Solutions

Additional Context

No response

[rhshadrach]

I'm positive on convert_dtypes supporting complex, currently it raises TypeError: complex128 cannot be converted to FloatingDtype if input has complex dtype. There is no Numpy-nullable complex array nor a PyArrow complex dtype, so for both backends this will be a noop (leaves complex data as-is).

convert_dtypes should be able to do this type of coercion from string complex looking

convert_dtypes does not convert strings to non-strings, so the current behavior of pd.DataFrame(['1.0+5j', '1.5-3j']).convert_dtypes() looks correct to me.