Description
Code Sample, a copy-pastable example if possible
import pandas as pd
df = pd.DataFrame({'B': [0, 1, 2, 4,3],'A':[7,4,6,9,3]},
index = [pd.Timestamp('20130101 09:00:00'),
pd.Timestamp('20130102 09:00:02'),
pd.Timestamp('20130103 09:00:03'),
pd.Timestamp('20130105 09:00:05'),
pd.Timestamp('20130106 09:00:06')])
print(df.corr())
'''
df.corr()
Out[53]:
B A
B 1.00000 0.19868
A 0.19868 1.00000
'''
df.rolling(window='3d').corr()
'''
B A
2013-01-01 09:00:00 B NaN NaN
A NaN NaN
2013-01-02 09:00:02 B 1.000000 -1.000000
A -1.000000 1.000000
2013-01-03 09:00:03 B 1.000000 -0.327327
A -0.327327 1.000000
2013-01-05 09:00:05 B 1.000000 0.609449
A 0.609449 1.000000
2013-01-06 09:00:06 B 1.000000 0.198680
A 0.198680 1.000000
'''Problem description
Due to some conflicts I'm not able to test it on pandas 0.25, so I'm not sure whether the problem is solved. You may find that the last coorelation value 0.198680 is exactly the same as perform corr() on the total dataframe. No matter how I change the timeoffset, namely the window parameter, the rolling(window='timeoffset string').corr() method returns cummulative correlation instead of correlation inside the window.
The reason why the expected output contains so many 'ones' is that rows which not fall in the time window should not be included into the calculation. For example, the 'ones' at the last row is calculated by the fourth row and the fifth row(the last row), because the indices is between 2013-01-03 09:00:06 and 2013-01-06 09:00:06 (the window parameter is 3d) and the correlation of these two point pairs is 1.
Expected Output
B A
2013-01-01 09:00:00 B NaN NaN
A NaN NaN
2013-01-02 09:00:02 B 1.000000 -1.000000
A -1.000000 1.000000
2013-01-03 09:00:03 B 1.000000 -0.327327
A -0.327327 1.000000
2013-01-05 09:00:05 B 1.000000 1.000000
A 1.000000 1.000000
2013-01-06 09:00:06 B 1.000000 1.000000
A 1.000000 1.000000
Output of pd.show_versions()
pandas: 0.24.2
pytest: 4.3.1
pip: 19.0.3
setuptools: 40.8.0
Cython: 0.29.6
numpy: 1.16.2
scipy: 1.2.1
pyarrow: None
xarray: None
IPython: 7.4.0
sphinx: 1.8.5
patsy: 0.5.1
dateutil: 2.8.0
pytz: 2018.9
blosc: None
bottleneck: 1.2.1
tables: 3.5.1
numexpr: 2.6.9
feather: None
matplotlib: 3.0.3
openpyxl: 2.6.1
xlrd: 1.2.0
xlwt: 1.3.0
xlsxwriter: 1.1.5
lxml.etree: 4.3.2
bs4: 4.7.1
html5lib: 1.0.1
sqlalchemy: 1.3.1
pymysql: None
psycopg2: None
jinja2: 2.10
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: None
gcsfs: None