ENH: Add Index.filter() method #51370
Conversation
| ) | ||
|
|
||
| if items is not None: | ||
| mask = [r in items for r in self] |
There was a problem hiding this comment.
This is super slow I guess? You should be able to use isin here.
There was a problem hiding this comment.
I just copied the pattern that is used in DataFrame.filter() :
Line 5531 in 4d74fbd
|
-1 here this is a confusing name (well it's an ok name except filter for dataframe does this) |
Yes, but it's a bit awkward, because the string accessors return boolean arrays. Comparison using example
I also think this PR will be more performant, because you are making just one pass through the index. |
Yes, but filter for |
|
Looks like this PR has gone stale and might need some discussion on an issue first before moving forward. Going to close for now, but we can reopen when ready to move forward |
pandas\pandas\tests\indexes\test_base.py:TestIndex.test_filter_string()pandas\pandas\tests\indexes\test_base.py:TestIndex.test_filter_int()doc/source/whatsnew/v2.0.0.rstfile if fixing a bug or adding a new feature.This is similar to
DataFrame.filter(), except it returns anIndexobject, and avoids any under-the-hood things that might be happening withDataFrame.filter()in terms of views/copies of theDataFrame. Some examples where this would be useful (and helpful when doing type checking).df.columns = [x for x in otherdf.columns if "at" in x], you can dodf.columns = otherdf.columns.filter(like="at")df.drop(columns=[x for x in df.columns if x.startswith("b")], you can dodf.drop(columns=df.columns.filter(regex=r"b.*")df.set_index([x for x in df.columns if x.endswith("z")]), you can dodf.set_index(df.columns.filter(r".*z$"))