FEAT: Add `Dataframe.Groupby.nth`

[JiaYaobo]

What do these changes do?

Add dataframe.groupby.nth , a counter part for https://pandas.pydata.org/pandas-docs/version/0.22/generated/pandas.core.groupby.GroupBy.nth.html

Very much in progress

Related issue number

Fixes https://github.com/xorbitsai/xorbits/issues/634

Check code requirements

• [x] tests added / passed (if needed)
• [x] Ensure all linting tests pass

[codecov[bot]]

Codecov Report

Merging #684 (fbb3200) into main (ebc391f) will increase coverage by 0.02%. The diff coverage is 95.49%.

@@            Coverage Diff             @@
##             main     #684      +/-   ##
==========================================
+ Coverage   93.53%   93.55%   +0.02%     
==========================================
  Files        1025     1026       +1     
  Lines       79405    79516     +111     
  Branches    16453    16475      +22     
==========================================
+ Hits        74271    74392     +121     
+ Misses       3453     3436      -17     
- Partials     1681     1688       +7     

Flag	Coverage Δ
unittests	93.44% <95.49%> (+0.02%)	:arrow_up:

Flags with carried forward coverage won't be shown. Click here to find out more.

Files Changed	Coverage Δ
python/xorbits/_mars/dataframe/groupby/nth.py	95.41% <95.41%> (ø)
python/xorbits/_mars/dataframe/groupby/__init__.py	96.55% <100.00%> (+0.12%)	:arrow_up:

... and 8 files with indirect coverage changes

[aresnow1]

Thanks for your contribution! Better to add tests to cover these changes.

[JiaYaobo]

Tests have beed added, tag you @aresnow1 here!

[JiaYaobo]

hmm, failures seem somewhat confused to me, it claims that "real shape" is (5, 3), however it should be (5, 4) since there are four columns, and CI ubuntu-latest, _mars/dataframe, 3.9 works fine, any help is appreciated! @aresnow1 @qinxuye

[aresnow1]

The failed check is for compatibility, you can install pandas v1.5.3 and run tests locally to reproduce it.

[JiaYaobo]

@aresnow1 thanks a lot! It seems that DataFrame.Groupby.nth has different behavoirs between versions. For example, with pandas==2.1.0

import pandas as pd
import numpy as np

df1 = pd.DataFrame(
    {
        "a": np.random.randint(0, 5, size=20),
        "b": np.random.randint(0, 5, size=20),
        "c": np.random.randint(0, 5, size=20),
        "d": np.random.randint(0, 5, size=20),
    }
)

df1.groupby("b").nth(0)

results:

However pandas == 1.5.3 results

So should I modify code to make it compatible between versions or just skip tests on pandas==1.5.3

[aresnow1]

@aresnow1 thanks a lot! It seems that DataFrame.Groupby.nth has different behavoirs between versions. For example, with pandas==2.1.0

import pandas as pd
import numpy as np

df1 = pd.DataFrame(
    {
        "a": np.random.randint(0, 5, size=20),
        "b": np.random.randint(0, 5, size=20),
        "c": np.random.randint(0, 5, size=20),
        "d": np.random.randint(0, 5, size=20),
    }
)

df1.groupby("b").nth(0)

results:

However pandas == 1.5.3 results

So should I modify code to make it compatible between versions or just skip tests on pandas==1.5.3

Let's skip the testing then. We will later change the minimum dependency for pandas to 2.0.

[JiaYaobo]

@aresnow1 Just skip tests when pandas version <= 1.5.3, and tests passed as expect, PTAL :)