BUG: .shift() on IntervalArray column raises ValueError

[datapythonista]

Code Sample, a copy-pastable example if possible

>>> import pandas
>>> interval_array = pandas.arrays.IntervalArray.from_arrays([1, 2, 3], [4, 5, 6])
>>> interval_array
IntervalArray([(1, 4], (2, 5], (3, 6]],
              closed='right',
              dtype='interval[int64]')
>>> interval_series = pandas.Series(interval_array)
>>> interval_series
0    (1, 4]
1    (2, 5]
2    (3, 6]
dtype: interval
>>> interval_series.shift(1)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "C:\msys64\home\UKC3153\src\pandas\pandas\core\series.py", line 3960, in shift
    fill_value=fill_value)
  File "C:\msys64\home\UKC3153\src\pandas\pandas\core\generic.py", line 9091, in shift
    fill_value=fill_value)
  File "C:\msys64\home\UKC3153\src\pandas\pandas\core\internals\managers.py", line 522, in shift
    return self.apply('shift', **kwargs)
  File "C:\msys64\home\UKC3153\src\pandas\pandas\core\internals\managers.py", line 395, in apply
    applied = getattr(b, f)(**kwargs)
  File "C:\msys64\home\UKC3153\src\pandas\pandas\core\internals\blocks.py", line 1828, in shift
    self.values.shift(periods=periods, fill_value=fill_value),
  File "C:\msys64\home\UKC3153\src\pandas\pandas\core\arrays\base.py", line 527, in shift
    dtype=self.dtype
  File "C:\msys64\home\UKC3153\src\pandas\pandas\core\arrays\interval.py", line 214, in _from_sequence
    return cls(scalars, dtype=dtype, copy=copy)
  File "C:\msys64\home\UKC3153\src\pandas\pandas\core\arrays\interval.py", line 159, in __new__
    verify_integrity=verify_integrity)
  File "C:\msys64\home\UKC3153\src\pandas\pandas\core\arrays\interval.py", line 177, in _simple_new
    left = left.astype(dtype.subtype)
  File "C:\msys64\home\UKC3153\src\pandas\pandas\core\indexes\numeric.py", line 331, in astype
    raise ValueError('Cannot convert NA to integer')
ValueError: Cannot convert NA to integer

Problem description

Applying a .shift() on a Series backed by the IntervalArray extension array raises an exception

Expected Output

0       NaN
1    (1, 4]
2    (2, 5]
dtype: interval

Output of pd.show_versions()

INSTALLED VERSIONS

python: 3.7.3.final.0
python-bits: 64
OS: Windows
OS-release: 10
machine: AMD64
processor: Intel64 Family 6 Model 142 Stepping 9, GenuineIntel
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
LOCALE: None.None

pandas: 0.25.0.dev0+592.gb563d452f
pytest: 4.5.0
pip: 19.1.1
setuptools: 41.0.1
Cython: 0.29.7
numpy: 1.16.3
scipy: 1.2.1
pyarrow: 0.11.1
xarray: 0.12.1
IPython: 7.5.0
sphinx: 2.0.1
patsy: 0.5.1
dateutil: 2.8.0
pytz: 2019.1
blosc: None
bottleneck: 1.2.1
tables: 3.5.1
numexpr: 2.6.9
feather: None
matplotlib: 3.0.3
openpyxl: 2.6.2
xlrd: 1.2.0
xlwt: 1.3.0
xlsxwriter: 1.1.8
lxml.etree: 4.3.3
bs4: 4.7.1
html5lib: 1.0.1
sqlalchemy: 1.3.3
pymysql: None
psycopg2: None
jinja2: 2.10.1
s3fs: None
fastparquet: 0.3.0
pandas_gbq: None
pandas_datareader: None
gcsfs: None

[jschendel]

Thanks, this is a dupe of #22428

This issue only occurs for dtypes that can't store NA, so a workaround in the meantime is to cast to a float subtype prior to the shift:

In [1]: import pandas

In [2]: interval_array = pandas.arrays.IntervalArray.from_arrays([1, 2, 3], [4, 5, 6])

In [3]: interval_series = pandas.Series(interval_array)

In [4]: interval_series
Out[4]: 
0    (1, 4]
1    (2, 5]
2    (3, 6]
dtype: interval

In [5]: interval_series.astype('interval[float64]').shift(1)
Out[5]: 
0           NaN
1    (1.0, 4.0]
2    (2.0, 5.0]
dtype: interval

Note: IntervalArray doesn't support IntNA subtypes yet, so there isn't a way to maintain integer endpoints when introducing NA.