Lookback methods (rolling/ewm/etc.) should accept series instead of just int/float

[JakeTheSnake3p0]

Issue: Using variable lookback-lengths in pandas methods

Per underlying stock asset, I utilize a homodyne discriminator derived from close prices which determines the appropriate lookback length for other indicators such as macd, smart volume, ehlers rsi, etc.

When I retrieve history on an asset, the data arrives in bulk. I want to apply my indicators en-masse to the entire dataset instead of having to iterate through each row in order to generate the appropriate values (lookback period, then indicator values). That means ideally I would create the 'lookback period' column on the entire dataset through a homodyne discriminator method/class and subsequently apply all of my indicator classes to the same dataset, most of which may depend on the lookback period.

Given the current implementation of ewm/rolling which seemingly accept only a static integer/float, all of my calculations will be off.

Ideal code which generates the error:

import pandas as pd
import numpy as np

# Initialize dataframe with source data
df = pd.DataFrame()
df['source'] = pd.Series(range(100, 200))

# Typically 'period' is generated by some class
# such as a homodyne discriminator for which
# other classes depend on for lookback length
df['period'] = np.nan
df.period.iloc[0] = 1
df.period.iloc[-1] = 10
df.period.interpolate(inplace=True)
df.period = df.period.astype('int64', copy=False)
print(df)

# A given class that depends on the lookback
# length determined by the discriminator
df['dependant'] = df.source.ewm(span=df.period, adjust=False).mean()
print(df)

Error:

ValueError: The truth value of a Series is ambiguous. Use a.empty, a.bool(), a.item(), a.any() or a.all().

INSTALLED VERSIONS

commit: None
python: 3.6.7.final.0
python-bits: 64
OS: Linux
OS-release: 4.4.0-1084-aws
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: en_US.UTF-8
LANG: en_US.UTF-8
LOCALE: en_US.UTF-8

pandas: 0.23.4
pytest: 4.2.0
pip: 19.0.1
setuptools: 40.8.0
Cython: 0.29.2
numpy: 1.14.5
scipy: 1.1.0
pyarrow: None
xarray: None
IPython: 7.2.0
sphinx: None
patsy: 0.5.1
dateutil: 2.7.5
pytz: 2018.9
blosc: None
bottleneck: None
tables: 3.4.4
numexpr: 2.6.8
feather: None
matplotlib: 3.0.2
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: None
lxml: None
bs4: None
html5lib: None
sqlalchemy: 1.2.16
pymysql: None
psycopg2: None
jinja2: 2.10
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: None
None

[TomAugspurger]

Looks like a duplicate of #23002