BUG: axis kw not propogating on pct_change #11150
Conversation
axis option was not correctly set on fillna
|
Thanks. Can you add a test for this? |
|
@Tux1 pls add a test for this |
|
Sorry. It is my first commit. |
|
pls see the contributing docs: http://pandas.pydata.org/pandas-docs/stable/contributing.html tests / fix should be on a single pr. |
|
Sorry... Moreover the test was wrong. |
| @@ -1851,6 +1851,15 @@ def test_pipe_panel(self): | |||
| with tm.assertRaises(ValueError): | |||
| result = wp.pipe((f, 'y'), x=1, y=1) | |||
|
|
|||
| def test_pct_change(self): | |||
| pnl = DataFrame([np.arange(0, 40, 10), np.arange(0, 40, 10), np.arange(0, 40, 10)]).astype(np.float64) | |||
|
pls add a whatsnew note for 0.17.1 (bug fix), use this PR number |
| pnl.iat[1,0] = np.nan | ||
| pnl.iat[1,1] = np.nan | ||
|
|
||
| expected = pnl.ffill(axis=1).pct_change(axis=1, fill_method=None) |
There was a problem hiding this comment.
test with axis 0, 2 as well (if their are existing tests for that, pls move them here)
further this should be a bit more generic (or append _frame) to the test name as the generic.py tests should test Series,DataFrame,Panel
|
@Tux1 can you rebase / update |
test regarding buf fix on pct_change propagation axis revamping test_pct_change. Rename, multi axis and error fixed
|
Sorry for several commits. I think that is okay now ? whatsnew edited, commit rebased |
| @@ -96,7 +96,7 @@ Bug Fixes | |||
|
|
|||
|
|
|||
|
|
|||
|
|
|||
| - Bug in ``DataFrame.pct_change()`` was not propagating axis on fillna method (:issue:`11150`) | |||
There was a problem hiding this comment.
use double backticks around axis (and say axis keyword) and 'on .fillna() method'
|
some comments, pls squash when finished |
test_pct_change with loop over axis backtick added
Add an optional argument to _NDFrameIndexer._align_series to indicate if the indexer is from a MultiIndex
|
pls rebase on master / squash, see docs here |
| def test_pct_change_frame(self): | ||
| pnl = DataFrame([np.arange(0, 40, 10), np.arange(0, 40, 10), np.arange(0, 40, 10)]).astype(np.float64) | ||
| pnl.iat[1,0] = np.nan | ||
| pnl.iat[1,1] = np.nan |
There was a problem hiding this comment.
I mean that this function needs to move to tests\test_frame.py (and not in test_generic.py)
|
Sorry, my repo is a mess... I will be better next time |
|
merged via 334b076 thanks! |
axis option was not correctly set on fillna in pct_change