Implement the missing Series.swaplevel #1919
Conversation
|
FYI: You have to remove the API from "missing" files and add a link in the doc. |
|
You can find the missing files in |
Codecov Report
@@ Coverage Diff @@
## master #1919 +/- ##
==========================================
+ Coverage 94.08% 94.16% +0.08%
==========================================
Files 41 41
Lines 10019 10022 +3
==========================================
+ Hits 9426 9437 +11
+ Misses 593 585 -8
Continue to review full report at Codecov.
|
databricks/koalas/series.py
Outdated
| index_map = list(zip(self._internal.index_spark_column_names, self._internal.index_names)) | ||
| index_map[i], index_map[j], = index_map[j], index_map[i] | ||
| index_spark_column_names, index_names = zip(*index_map) | ||
| internal = self._kdf._internal.copy( |
There was a problem hiding this comment.
self._internal instead ofself._kdf._internal ?
There was a problem hiding this comment.
May I ask what's their difference at usage?
I see return self._kdf._internal.select_column(self._column_label) is for self._internal.
There was a problem hiding this comment.
self._kdf._internal contains other columns from the anchor DataFrame.
E.g.,:
>>> midx = pd.MultiIndex.from_arrays([['a', 'b'], [1, 2]], names = ['word', 'number'])
>>> kdf = ks.DataFrame({'a': [1, 2], 'b': ['x', 'y']}, index=midx)
>>> kdf
a b
word number
a 1 1 x
b 2 2 y
>>> kdf.b.swaplevel()
number word
1 a 1
2 b 2
Name: a, dtype: int64should be:
>>> kdf.b.swaplevel()
number word
1 a x
2 b y
Name: b, dtype: objectThere was a problem hiding this comment.
Thanks for explaining!
I'm a little confused by "contains". Their spark_frame seems to be identical.
>>> kdf.b._internal.spark_frame.show()
+-----------------+-----------------+---+---+-----------------+
|__index_level_0__|__index_level_1__| a| b|__natural_order__|
+-----------------+-----------------+---+---+-----------------+
| a| 1| 1| x| 42949672960|
| b| 2| 2| y| 94489280512|
+-----------------+-----------------+---+---+-----------------+
>>> kdf.b._kdf._internal.spark_frame.show()
+-----------------+-----------------+---+---+-----------------+
|__index_level_0__|__index_level_1__| a| b|__natural_order__|
+-----------------+-----------------+---+---+-----------------+
| a| 1| 1| x| 42949672960|
| b| 2| 2| y| 94489280512|
+-----------------+-----------------+---+---+-----------------+
There was a problem hiding this comment.
I mean their metadata:
>>> kdf.b._internal.column_labels
[('b',)]
>>> kdf.b._kdf._internal.column_labels
[('a',), ('b',)]or
>>> kdf.b._internal.data_spark_columns
[Column<b'b'>]
>>> kdf.b._kdf._internal.data_spark_columns
[Column<b'a'>, Column<b'b'>]The column_labels or data_spark_columns in the InternalFrame for Series contains only one for the Series.
There was a problem hiding this comment.
I see, that's clear to me now. Thank you!
No description provided.