Calling arbitrary function from pandas rolling series-CodePudding

I can call an arbitrary method defined via a string variable on a pandas series. The way I would do it would be like this:

import pandas as pd

method_name = 'mean'
pd.Series([1, 2, 3]).__getattr__(method_name)()

Now I want to do the same with a rolling pandas series, the way I would do it would be like this:

import pandas as pd

method_name = 'mean'
pd.Series([1, 2, 3]).rolling(window=1).__getattr__(method_name)()

I get the following error when doing it:

AttributeError: 'Rolling' object has no attribute 'mean'

Is there a way to call an arbitrary method (mean, median, max, min, quantile) on a rolling pandas series?

Thanks!

CodePudding user response：

You can read about getattr and getattribute from the Python documentation (Data Model): getattr and getattribute

The use of getattribute will solve your problem;

import pandas as pd

method_name = 'mean'
pd.Series([1, 2, 3]).rolling(window=1).__getattribute__(method_name)()

Output

0    1.0
1    2.0
2    3.0
dtype: float64

CodePudding user response：

The method __getattr__ must be implemented for each object. This is unlike __getatrribute__ which is a method in the base object class in Python. When calling __getatrr__, you will get that object implementation of this method (if any). For rolling, the definition is below, which basically does some checking for caching/pickling purposes (the _internal_names_set), if not found, it checks the dataframe (the self.obj):

def __getattr__(self, attr: str):
    if attr in self._internal_names_set:
        return object.__getattribute__(self, attr)
    if attr in self.obj:
        return self[attr]

    raise AttributeError(
        f"'{type(self).__name__}' object has no attribute '{attr}'"
    )

To avoid this, use the getattr builtin function:

>> getattr(df.rolling(3), 'mean')

I like getattr because it allows you to specify a third, optional argument for when the attribute is not defined:

>> getattr(df.rolling(3), 'mean', None)