Implement min, max, sum and mean of Rolling in Series and DataFrame

[itholic]

This PR implement min, max, sum and mean of Rolling in Series and DataFrame

For Series:

>>> kser = ks.Series([2, 3, float("nan"), 10])
>>> kser
0     2.0
1     3.0
2     NaN
3    10.0
Name: 0, dtype: float64

>>> ksr.rolling(2).sum()
0    NaN
1    5.0
2    3.0
3    NaN
Name: 0, dtype: float64

>>> ks.rolling(2).min()
0    NaN
1    2.0
2    3.0
3    NaN
Name: 0, dtype: float64

>>> ks.rolling(2).max()
0    NaN
1    3.0
2    3.0
3    NaN
Name: 0, dtype: float64

>>> ks.rolling(2).mean()
0    NaN
1    2.5
2    3.0
3    NaN
Name: 0, dtype: float64

For DataFrame

>>> kdf = ks.DataFrame({'a': [1, float('nan'), 3], 'b': [1.0, 2.0, 3.0]})
>>> kdf
     a    b
0  1.0  1.0
1  NaN  2.0
2  3.0  3.0

>>> kdf.rolling(2).sum()
     a    b
0  NaN  NaN
1  1.0  3.0
2  NaN  5.0

>>> kdf.rolling(2).min()
     a    b
0  NaN  NaN
1  1.0  1.0
2  NaN  2.0

>>> kdf.rolling(2).max()
     a    b
0  NaN  NaN
1  1.0  2.0
2  NaN  3.0

>>> kdf.rolling(2).mean()
     a    b
0  NaN  NaN
1  1.0  1.5
2  NaN  2.5

relates to #977

[codecov-io]

Codecov Report

❗ No coverage uploaded for pull request base (master@9434e48). Click here to learn what that means.
The diff coverage is 86.2%.

@@            Coverage Diff            @@
##             master     #996   +/-   ##
=========================================
  Coverage          ?   92.46%           
=========================================
  Files             ?       34           
  Lines             ?     6698           
  Branches          ?        0           
=========================================
  Hits              ?     6193           
  Misses            ?      505           
  Partials          ?        0

Impacted Files	Coverage Δ
databricks/koalas/generic.py	95.73% <ø> (ø)
databricks/koalas/window.py	93.03% <86.2%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 9434e48...364254c. Read the comment docs.

[HyukjinKwon]

Can we move this into _apply_as_series_or_frame and use self._window? We can match the implementation into Expanding one I think.

[itholic]

@HyukjinKwon
Thanks for the comment!

but maybe is it possible although this window and self._window are used for different purpose like below?

i think to calculate F.lag(scol, self._window_val).over(window) >= self._min_periods, we need another window rather than self._window

[HyukjinKwon]

Looks fine otherwise.

[HyukjinKwon]

@itholic can you rebase please?

[itholic]

@HyukjinKwon sure, i just did.

[softagram-bot]

Softagram Impact Report for pull/996 (head commit: 364254c)

⭐ Change Overview

(Open in Softagram Desktop for full details)

📄 Full report

• Permalink: Full report for pull/996

Impact Report explained. Give feedback on this report to [email protected]

[HyukjinKwon]

Let me clean up by myself. merged

[itholic]

@HyukjinKwon Thanks :)