API: Period and PeriodIndex do not support some offset aliases

[agraboso]

Period and PeriodIndex raise ValueError when freq is any of the following strings:

'BM', 'CBM', 'MS', 'BMS', 'CBMS', 'BQ', 'QS', 'BQS', 'BA', 'AS', 'BAS', 'BH'

date_range is fine in all cases. Also confirmed on master.

Possible related to #4878.

Code Sample, a copy-pastable example if possible

import pandas as pd


FREQS = ['B', 'C', 'D', 'W', 'M', 'BM', 'CBM', 'MS', 'BMS', 'CBMS', 'Q',
         'BQ', 'QS', 'BQS', 'A', 'BA', 'AS', 'BAS', 'BH', 'H', 'T', 'min',
         'S', 'L', 'U', 'N']


for freq in FREQS:
    try:
        pd.Period('2016-01-01', freq=freq)
    except ValueError:
        print('Cannot create Period with freq={}'.format(freq))

    p = pd.Period('2016-01-01', freq='N')
    try:
        p.asfreq(freq)
    except ValueError:
        print('Cannot modify Period to freq={}'.format(freq))

    try:
        pd.PeriodIndex(start='2016-01-01', periods=10, freq=freq)
    except ValueError:
        print('Cannot create PeriodIndex with freq={}'.format(freq))

    pi = pd.PeriodIndex(start='2016-01-01', periods=10, freq='N')
    try:
        pi.asfreq(freq)
    except ValueError:
        print('Cannot modify PeriodIndex to freq={}'.format(freq))

    try:
        pd.date_range(start='2016-01-01', periods=10, freq=freq)
    except ValueError:
        print('Cannot create date_range with freq={}'.format(freq))

Output

Cannot create Period with freq=BM
Cannot modify Period to freq=BM
Cannot create PeriodIndex with freq=BM
Cannot modify PeriodIndex to freq=BM
Cannot create Period with freq=CBM
Cannot modify Period to freq=CBM
Cannot create PeriodIndex with freq=CBM
Cannot modify PeriodIndex to freq=CBM
Cannot create Period with freq=MS
Cannot modify Period to freq=MS
Cannot create PeriodIndex with freq=MS
Cannot modify PeriodIndex to freq=MS
Cannot create Period with freq=BMS
Cannot modify Period to freq=BMS
Cannot create PeriodIndex with freq=BMS
Cannot modify PeriodIndex to freq=BMS
Cannot create Period with freq=CBMS
Cannot modify Period to freq=CBMS
Cannot create PeriodIndex with freq=CBMS
Cannot modify PeriodIndex to freq=CBMS
Cannot create Period with freq=BQ
Cannot modify Period to freq=BQ
Cannot create PeriodIndex with freq=BQ
Cannot modify PeriodIndex to freq=BQ
Cannot create Period with freq=QS
Cannot modify Period to freq=QS
Cannot create PeriodIndex with freq=QS
Cannot modify PeriodIndex to freq=QS
Cannot create Period with freq=BQS
Cannot modify Period to freq=BQS
Cannot create PeriodIndex with freq=BQS
Cannot modify PeriodIndex to freq=BQS
Cannot create Period with freq=BA
Cannot modify Period to freq=BA
Cannot create PeriodIndex with freq=BA
Cannot modify PeriodIndex to freq=BA
Cannot create Period with freq=AS
Cannot modify Period to freq=AS
Cannot create PeriodIndex with freq=AS
Cannot modify PeriodIndex to freq=AS
Cannot create Period with freq=BAS
Cannot modify Period to freq=BAS
Cannot create PeriodIndex with freq=BAS
Cannot modify PeriodIndex to freq=BAS
Cannot create Period with freq=BH
Cannot modify Period to freq=BH
Cannot create PeriodIndex with freq=BH
Cannot modify PeriodIndex to freq=BH

output of pd.show_versions()

INSTALLED VERSIONS
------------------
commit: None
python: 3.5.2.final.0
python-bits: 64
OS: Darwin
OS-release: 15.5.0
machine: x86_64
processor: i386
byteorder: little
LC_ALL: en_US.UTF-8
LANG: en_US.UTF-8

pandas: 0.18.1
nose: 1.3.7
pip: 8.1.2
setuptools: 23.0.0
Cython: 0.24
numpy: 1.11.0
scipy: 0.17.1
statsmodels: 0.6.1
xarray: None
IPython: 4.2.0
sphinx: None
patsy: 0.4.1
dateutil: 2.5.3
pytz: 2016.4
blosc: None
bottleneck: None
tables: None
numexpr: 2.6.0
matplotlib: 1.5.1
openpyxl: None
xlrd: 1.0.0
xlwt: None
xlsxwriter: None
lxml: 3.6.0
bs4: 4.4.1
html5lib: None
httplib2: None
apiclient: None
sqlalchemy: 1.0.13
pymysql: None
psycopg2: None
jinja2: 2.8
boto: None
pandas_datareader: None
None

[jreback]

@agraboso

with the exception of BH I don't think any of those make sense. They are representing a point-in-time, e.g. a month-START, rather than a Period which is a timespan.

@sinhrks

[jreback]

actually, let's repurpose this issue to provide a more informative error for why these are not supported. E.g. they are valid point-in-time aliases (and not just a random string), so we can check for and report better.

[sinhrks]

Better error report is one thing, or it may be useful to support it (for DatetimeIndex compat).

I think monthly freq like BusinessMonthEnd can use the same internal repr as MonthEnd, and needs small changes in some impl (like to_timestamp).

[EricLeer]

take

[simonjayhawkins]

@jreback I'm changing the title back to something like the original because of subsequent comment #13871 (comment)

[simonjayhawkins]

and closing #4878 as duplicate

[simonjayhawkins]

d = pd.date_range(start="2016-01-01", periods=10, freq="BM")
print(d)
print(d.to_period())
pd.period_range(start="2016-01-01", periods=10, freq="BM")

DatetimeIndex(['2016-01-29', '2016-02-29', '2016-03-31', '2016-04-29',
               '2016-05-31', '2016-06-30', '2016-07-29', '2016-08-31',
               '2016-09-30', '2016-10-31'],
              dtype='datetime64[ns]', freq='BM')
PeriodIndex(['2016-01', '2016-02', '2016-03', '2016-04', '2016-05', '2016-06',
             '2016-07', '2016-08', '2016-09', '2016-10'],
            dtype='period[M]')
---------------------------------------------------------------------------
...
ValueError: Invalid frequency: <BusinessMonthEnd>

Better error report is one thing, or it may be useful to support it (for DatetimeIndex compat).

I think monthly freq like BusinessMonthEnd can use the same internal repr as MonthEnd, and needs small changes in some impl (like to_timestamp).

Since to_period() works for a DatetimeIndex with freq='BM', it may make sense to use the same logic used to convert these and accept them in the period_range, Period, PeriodIndex constructors.