groupby nth for multiindex regression

[hayd]

regression in groupby nth, it raises an index error on MI in master. This is my fault (need to add tests, should also test for axis!=0).

http://stackoverflow.com/questions/22361385/integer-position-based-indexing-of-multi-index-dataframes

reason this raise is as I use obj.loc[[[]] :s

[jreback]

really? I wrote that answer on master

example?

[hayd]

Ah, think difference is using manual index rather than levels kwarg:

df = pd.DataFrame({'color': {0: 'green', 1: 'green', 2: 'red', 3: 'red', 4: 'red'}, 'food': {0: 'ham', 1: 'eggs', 2: 'eggs', 3: 'ham', 4: 'pork'}, 'two': {0: 1.5456590000000001, 1: -0.070345000000000005, 2: -2.4004539999999999, 3: 0.46206000000000003, 4: 0.52350799999999997}, 'one': {0: 0.56573799999999996, 1: -0.9742360000000001, 2: 1.033801, 3: -0.78543499999999999, 4: 0.70422799999999997}}).set_index(['color', 'food'])

In [8]: df.groupby(level=0).nth(2)
Out[8]: 
                 one       two
color food                    
red   pork  0.704228  0.523508

In [9]: df.groupby(level=0).nth(3)
IndexError: index out of bounds

[hayd]

It's possible this is a one char change (to use iloc rather than loc).

Would be good to fix on axis behaviour here (clearly there was no coverage before and I don't think it worked).

[jreback]

@hayd you doing for 0.14?

[hayd]

Should be in 0.14, I'll have some more time to do all these at the end of the week / next week. Very behind with all these PRs (should've been doing them incrementally rather than saving them all up)! :s

[jreback]

hahah...ok np

[jreback]

@hayd ping!

[jreback]

@hayd do you have something for this? (or is this covered already?)

is this a regression?

[jreback]

@hayd ?

[jreback]

yeh..this is easy....i'll put up a PR