0.17 changes behavior of division of DataFrame by Series in presence of DateTimeIndex

[twiecki]

Older versions up to 0.16.2:

import pandas as pd
print(pd.__version__)
index = pd.date_range(start='2000-1-1', end='2000-1-3')
pd.DataFrame({1: [3, 3, 3], 2: [6, 6, 6]}, index=index) / pd.Series([2, 2, 2], index=index)

0.14.0rc1-33-g8c1f550
1   2
2000-01-01  1.5 3
2000-01-02  1.5 3
2000-01-03  1.5 3

0.17:

0.17.0
2000-01-01 00:00:00 2000-01-02 00:00:00 2000-01-03 00:00:00 1   2
2000-01-01  NaN NaN NaN NaN NaN
2000-01-02  NaN NaN NaN NaN NaN
2000-01-03  NaN NaN NaN NaN NaN

Work-around:
Use df.divide(series, axis='index').

Is this change in API intentional? Couldn't find it in the release announcements.

[TomAugspurger]

Looks like this deprecation on automatic time-series broadcasting, from #2304

[jreback]

@twiecki this is as expected and has a specific mention in the whatsnew here

this was a DeprecationWarning since 0.8.0 FYI

[twiecki]

Thanks @jreback. Never noticed the warning.

[jreback]

@twiecki yep. It has been around for quite a long time. Finally got around to removing it :)

[twiecki]

I have full confidence that it was the right move :).