DOC: better warning on chained resample(...)

[chris-b1]

Code Sample, a copy-pastable example if possible

In [49]: df = pd.DataFrame({'a': 1.}, index=pd.date_range('2014-01-01', '2015-12-31'))

In [50]: df.resample('M').pct_change()
FutureWarning: .resample() is now a deferred operation
use .resample(...).mean() instead of .resample(...)
  if __name__ == '__main__':
Out[50]: 
              a
2014-01-31  NaN
<snip>

Expected Output

no warning

output of pd.show_versions()

INSTALLED VERSIONS
------------------
commit: None
python: 2.7.11.final.0
python-bits: 64
OS: Windows
OS-release: 7
machine: AMD64
processor: Intel64 Family 6 Model 58 Stepping 9, GenuineIntel
byteorder: little
LC_ALL: None
LANG: None

pandas: 0.18.1
nose: 1.3.7
pip: 8.1.0
setuptools: 20.2.2
Cython: 0.22.1
numpy: 1.10.4
scipy: 0.16.0
statsmodels: 0.6.1
xarray: None
IPython: 4.0.3
sphinx: 1.2.3
patsy: 0.3.0
dateutil: 2.4.2
pytz: 2015.7
blosc: None
bottleneck: 1.0.0
tables: 3.2.2
numexpr: 2.5
matplotlib: 1.5.1
openpyxl: 2.0.2
xlrd: 0.9.3
xlwt: 1.0.0
xlsxwriter: 0.7.3
lxml: 3.4.4
bs4: 4.3.2
html5lib: 0.999
httplib2: 0.9.2
apiclient: 1.5.0
sqlalchemy: 1.0.5
pymysql: None
psycopg2: None
jinja2: 2.8
boto: 2.38.0
pandas_datareader: 0.2.0

[chris-b1]

Actually, this isn't spurious - what I meant to write is this. I'll leave this open - I think the warning could be better, to make it more clear that the mean of the data is being taken.

df.resample('M').last().pct_change()

[jorisvandenbossche]

Do you have a suggestion for a better warning? (PR's welcome :-))

df.resample('M').last().pct_change()

It's not fully clear what you mean with this

[chris-b1]

I'll do a PR, what I want it to say is (in essence, hopefully in less words) -

.resample() is a deferred operation. You used that deferred object as if it were a {klass} - this materializes the deferred object by taking the mean() of each date group, which is deprecated behavior. Update your code <...>

What I was actually trying to do is take the month over month percent change - which is what .last().pct_change() does. But I forgot the last, so it took the mean of month, then the pct_change of those means.

[jreback]

yes this is mainly for back-compat.

basically in < 0.18.0 you should have been doing

df.resample('M', how='mean').pct_change()

and so in >= 0.18.0 you implicity get this. It IS correct actually. I think the warning could also include the deferred op reference (e.g. pct_change()) I suppose, BUT the warning actually comes out before.

You might have a look at Resampler.plot().

[jreback]

thanks!