pandas.rolling.max() shut down reopen #24218

kid3night · 2018-12-11T02:23:58Z

Code Sample, a copy-pastable example if possible

# Your code here
import pandas as pd

a = [0, 100900, 101000, 101000, 101000, 100900, 100900, 100900, 
	 100900, 100900, 100900, 100900, 100900, 100900, 100900, 
	 100900, 100900, 100900, 100900, 100900, 0, 100900, 100900, 
	 100900, 100900, 100900, 100900, 0, 0, 0, 101000, 0, 0, 0, 
	 101000, 100900, 100900, 100900, 100900, 100900, 100900, 
	 100900, 0, 101000, 101000, 101000, 101000, 0, 0, 0, 100900, 
	 100900, 100900, 100900, 100900, 100900, 100900, 100900, 
	 100900, 100900]

ix = pd.DatetimeIndex(['2018-06-01 09:33:49', '2018-06-01 09:33:50',
	               '2018-06-01 09:33:50', '2018-06-01 09:33:52',
	               '2018-06-01 09:33:52', '2018-06-01 09:33:53',
	               '2018-06-01 09:33:53', '2018-06-01 09:33:53',
	               '2018-06-01 09:33:53', '2018-06-01 09:33:53',
	               '2018-06-01 09:33:53', '2018-06-01 09:33:53',
	               '2018-06-01 09:33:53', '2018-06-01 09:33:53',
	               '2018-06-01 09:33:53', '2018-06-01 09:33:53',
	               '2018-06-01 09:33:53', '2018-06-01 09:33:53',
	               '2018-06-01 09:33:53', '2018-06-01 09:33:53',
	               '2018-06-01 09:33:53', '2018-06-01 09:33:53',
	               '2018-06-01 09:33:53', '2018-06-01 09:33:53',
	               '2018-06-01 09:33:53', '2018-06-01 09:33:53',
	               '2018-06-01 09:33:53', '2018-06-01 09:33:54',
	               '2018-06-01 09:33:54', '2018-06-01 09:33:54',
	               '2018-06-01 09:33:54', '2018-06-01 09:33:54',
	               '2018-06-01 09:33:55', '2018-06-01 09:33:55',
	               '2018-06-01 09:33:55', '2018-06-01 09:33:55',
	               '2018-06-01 09:33:55', '2018-06-01 09:33:55',
	               '2018-06-01 09:33:55', '2018-06-01 09:33:55',
	               '2018-06-01 09:33:55', '2018-06-01 09:33:55',
	               '2018-06-01 09:33:55', '2018-06-01 09:33:56',
	               '2018-06-01 09:33:56', '2018-06-01 09:33:56',
	               '2018-06-01 09:33:57', '2018-06-01 09:33:57',
	               '2018-06-01 09:33:57', '2018-06-01 09:33:57',
	               '2018-06-01 09:33:58', '2018-06-01 09:33:58',
	               '2018-06-01 09:33:58', '2018-06-01 09:33:58',
	               '2018-06-01 09:33:58', '2018-06-01 09:33:58',
	               '2018-06-01 09:33:58', '2018-06-01 09:33:58',
	               '2018-06-01 09:33:58', '2018-06-01 09:33:58'],
	              dtype='datetime64[ns]', name='0', freq=None)

series = pd.Series(a, index=ix)
series.rolling('3S', closed='left').max()

Problem description

I got a sample data and want to use it to get the rolling max, min, sum etc.
But every time I did rolling max/min on it, the python kernal would shut down and shows 'python.exe has stopped working'

Expected Output

Output of `pd.show_versions()`

[paste the output of pd.show_versions() here below this line]

INSTALLED VERSIONS

commit: None
python: 3.6.5.final.0
python-bits: 64
OS: Windows
OS-release: 10
machine: AMD64
processor: Intel64 Family 6 Model 60 Stepping 3, GenuineIntel
byteorder: little
LC_ALL: None
LANG: None
LOCALE: None.None

pandas: 0.23.0
pytest: 3.5.1
pip: 10.0.1
setuptools: 39.1.0
Cython: 0.28.2
numpy: 1.14.3
scipy: 1.1.0
pyarrow: None
xarray: None
IPython: 6.4.0
sphinx: 1.7.4
patsy: 0.5.0
dateutil: 2.7.3
pytz: 2018.4
blosc: None
bottleneck: 1.2.1
tables: 3.4.3
numexpr: 2.6.5
feather: None
matplotlib: 2.2.2
openpyxl: 2.5.3
xlrd: 1.1.0
xlwt: 1.3.0
xlsxwriter: 1.0.4
lxml: 4.2.1
bs4: 4.6.0
html5lib: 1.0.1
sqlalchemy: 1.2.7
pymysql: None
psycopg2: None
jinja2: 2.10
s3fs: 0.1.6
fastparquet: None
pandas_gbq: None
pandas_datareader: None

The text was updated successfully, but these errors were encountered:

WillAyd · 2018-12-11T05:43:12Z

If you could make your code self-contained and copy/pastable would make it much easier for someone to take a look

kid3night · 2018-12-11T08:03:46Z

If you could make your code self-contained and copy/pastable would make it much easier for someone to take a look

No problem, I have updated it.

kid3night · 2018-12-14T01:17:17Z

If you could make your code self-contained and copy/pastable would make it much easier for someone to take a look

Is there any info needed right now?

WillAyd · 2018-12-14T02:18:07Z

Have you tried on master? IIRC we fixed something similar for windows recently

kid3night · 2018-12-14T05:14:27Z

Have you tried on master? IIRC we fixed something similar for windows recently

I just updated and tried again. The problem still exist.

WillAyd · 2018-12-14T05:17:21Z

Hmm I couldn't reproduce so likely a Windows issue. Let's see if someone else can

kid3night · 2018-12-14T05:20:41Z

Hmm I couldn't reproduce so likely a Windows issue. Let's see if someone else can
Yea, I tried the same thing one my Mac and the problem doesn't exist.

kid3night · 2018-12-19T06:50:57Z

Hmm I couldn't reproduce so likely a Windows issue. Let's see if someone else can

I want update one info:
If I use delete the closed='left', then the code will not crash. It seems that the problem exist in the boundary part.

timhunderwood · 2019-01-05T10:40:25Z

Hello,

I was able to reproduce this on Windows.

From some debugging, the issue occurs when you get to:

pandas.core.window._Rolling#_apply

When it tries to call:

return cfunc(arg, window, minp, indexi, closed, **kwargs)

with cfunc refering to "roll_max".

I think something goes wrong in:
pandas._libs.window#_roll_min_max

I can reproduce the issue by running the below:

import pandas._libs.window
import numpy

values = numpy.linspace(0, 1, 5)
window = 2
minp = 1
index = numpy.array(range(5)).astype(numpy.int64)
closed = "left"

pandas._libs.window.roll_max(values, window, minp, index, closed)

The issue does not occur if closed="right" or closed="both" or closed=None.

Let me know if I can help debug/resolve.

Output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.7.2.final.0
python-bits: 64
OS: Windows
OS-release: 10
machine: AMD64
processor: Intel64 Family 6 Model 94 Stepping 3, GenuineIntel
byteorder: little
LC_ALL: None
LANG: None
LOCALE: None.None
pandas: 0.23.4
pytest: None
pip: 18.1
setuptools: 40.6.3
Cython: None
numpy: 1.15.4
scipy: 1.2.0
pyarrow: None
xarray: None
IPython: None
sphinx: None
patsy: None
dateutil: 2.7.5
pytz: 2018.7
blosc: None
bottleneck: None
tables: None
numexpr: None
feather: None
matplotlib: None
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: None
lxml: None
bs4: None
html5lib: None
sqlalchemy: None
pymysql: None
psycopg2: None
jinja2: None
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: None

WillAyd · 2019-01-05T16:58:27Z

@timhunderwood investigation and PRs are always welcome!

timhunderwood · 2019-01-06T17:55:52Z

When I build version 0.24 locally ('0.24.0.dev0+1522.gd106e9975') this issue doesn't occur. Looks like it has already been fixed between 0.23.4 and now.

jreback · 2019-01-06T19:25:08Z

@timhunderwood you are running on windows? if want to put up a PR with a validation test (that cycles thru all of the closed options), would be great.

timhunderwood · 2019-01-06T19:53:44Z

hi @jreback , yes, I am testing the above on windows. I will try and make a PR to test all the options for closed.

timhunderwood · 2019-01-08T21:58:12Z

I had a look at pandas/tests/test_window.py - I think this issue is a duplicate of #21704 and was fixed in #21853. This PR also added pandas.tests.test_window.TestRolling#test_closed_min_max_datetime which tests all the combinations of closed and should cover the above case.

Hope that helps - let me know if you think there is something that should be added to those tests.

jreback · 2019-01-08T22:34:33Z

thanks

WillAyd added the Needs Info Clarification about behavior needed to assess issue label Dec 11, 2018

WillAyd added Can't Repro and removed Needs Info Clarification about behavior needed to assess issue labels Dec 14, 2018

WillAyd added Windows Windows OS Window rolling, ewma, expanding and removed Can't Repro labels Jan 5, 2019

WillAyd added this to the Contributions Welcome milestone Jan 5, 2019

jreback closed this as completed Jan 8, 2019

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

pandas.rolling.max() shut down reopen #24218

pandas.rolling.max() shut down reopen #24218

kid3night commented Dec 11, 2018 •

edited

Loading

INSTALLED VERSIONS

WillAyd commented Dec 11, 2018

kid3night commented Dec 11, 2018

kid3night commented Dec 14, 2018

WillAyd commented Dec 14, 2018

kid3night commented Dec 14, 2018

WillAyd commented Dec 14, 2018

kid3night commented Dec 14, 2018

kid3night commented Dec 19, 2018

timhunderwood commented Jan 5, 2019

INSTALLED VERSIONS

WillAyd commented Jan 5, 2019

timhunderwood commented Jan 6, 2019

jreback commented Jan 6, 2019

timhunderwood commented Jan 6, 2019

timhunderwood commented Jan 8, 2019

jreback commented Jan 8, 2019

pandas.rolling.max() shut down reopen #24218

pandas.rolling.max() shut down reopen #24218

Comments

kid3night commented Dec 11, 2018 • edited Loading

Code Sample, a copy-pastable example if possible

Problem description

Expected Output

Output of pd.show_versions()

INSTALLED VERSIONS

WillAyd commented Dec 11, 2018

kid3night commented Dec 11, 2018

kid3night commented Dec 14, 2018

WillAyd commented Dec 14, 2018

kid3night commented Dec 14, 2018

WillAyd commented Dec 14, 2018

kid3night commented Dec 14, 2018

kid3night commented Dec 19, 2018

timhunderwood commented Jan 5, 2019

INSTALLED VERSIONS

WillAyd commented Jan 5, 2019

timhunderwood commented Jan 6, 2019

jreback commented Jan 6, 2019

timhunderwood commented Jan 6, 2019

timhunderwood commented Jan 8, 2019

jreback commented Jan 8, 2019

kid3night commented Dec 11, 2018 •

edited

Loading

Output of `pd.show_versions()`