Series.convert_dtypes(infer_objects=True, convert_string=True, convert_integer=True, convert_boolean=True, convert_floating=True, dtype_backend=None)[source]#

Convert columns to the best possible nullable dtypes.

If the dtype is numeric, and consists of all integers, convert to an appropriate integer extension type. Otherwise, convert to an appropriate floating type.

All other dtypes are always returned as-is as all dtypes in cudf are nullable.