BUG: window function count should count anything non-null · Issue #12541 · pandas-dev/pandas (original) (raw)
Right now np.Inf and non-numeric types are not counted.
Code Sample, a copy-pastable example if possible
import pandas as pd
import numpy as np
from pandas import DataFrame
df = DataFrame({'A': pd.date_range('20130101',periods=4), 'B': [0, 1, 2, np.Inf], 'C':['a', 'b', 'c', 'd']})
roll = df.rolling(window=2)
res = roll.count()
print(res)
## partial output A, B don't error
A B
0 1.0 1.0
1 2.0 2.0
2 2.0 2.0
3 2.0 1.0
## column C errors
Expected Output
A B C
0 1 1 1
1 2 2 2
2 2 2 2
3 2 2 2
output of pd.show_versions()
INSTALLED VERSIONS
------------------
commit: a0aaad98e8c567a3a04aef165cdb159fbd98d13f
python: 3.5.1.final.0
python-bits: 64
OS: Linux
OS-release: 3.13.0-76-generic
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
pandas: 0.18.0rc1+77.ga0aaad9.dirty
nose: 1.3.7
pip: 8.0.1
setuptools: 19.4
Cython: 0.23.4
numpy: 1.10.2
scipy: None
statsmodels: None
xarray: None
IPython: 4.0.0
sphinx: None
patsy: None
dateutil: 2.4.2
pytz: 2015.7
blosc: None
bottleneck: 1.0.0
tables: None
numexpr: None
matplotlib: None
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: None
lxml: None
bs4: None
html5lib: None
httplib2: None
apiclient: None
sqlalchemy: 1.0.10
pymysql: None
psycopg2: None
jinja2: None
boto: None
I think it can be patched, together with #12538, by
--- a/pandas/core/window.py
+++ b/pandas/core/window.py
@@ -509,19 +509,9 @@ class _Rolling_and_Expanding(_Rolling):
blocks, obj = self._create_blocks(how=None)
results = []
for b in blocks:
-
- if com.needs_i8_conversion(b.values):
- result = b.notnull().astype(int)
- else:
- try:
- result = np.isfinite(b).astype(float)
- except TypeError:
- result = np.isfinite(b.astype(float)).astype(float)
-
- result[pd.isnull(result)] = 0
-
- result = self._constructor(result, window=window, min_perio
- center=self.center).sum()
+ result = self._constructor(
+ b.notnull(), window=window, min_periods=0, center=self.
+ ).sum().astype(int)