Python class derived from pandas DataFrame with list/DataFrame attribute

Question

I am trying to create a class derived from pandas DataFrame. The class shall have each an attribute of type string, DataFrame and list. There is no problem with the string attribute, but DataFrame and list each cause a warning. Despite the warnings, the code seems to behave correctly.

Can anyone help me to fix my code? Or suppress the warnings?

Code:

import pandas as pd

class MyClass(pd.DataFrame):
    def __init__(self, arg_string, arg_dataframe, *arg_params):
        pd.DataFrame.__init__(self)
        self._string    = arg_string
        self._dataframe = arg_dataframe
        self._params    = arg_params

if __name__=='__main__':
    df = pd.DataFrame()
    c1 = MyClass("test", df, 1, 2, 3)

    print(c1._string)
    print(c1._dataframe)
    print(c1._params)

Warning message:

so_example.py:7: UserWarning: Pandas doesn't allow columns to be created via a new attribu
te name - see https://pandas.pydata.org/pandas-docs/stable/indexing.html#attribute-access
  self._dataframe = arg_dataframe

so_example.py:8: UserWarning: Pandas doesn't allow columns to be created via a new attribu
te name - see https://pandas.pydata.org/pandas-docs/stable/indexing.html#attribute-access
  self._params    = arg_params

Stdout:

test
Empty DataFrame
Columns: []
Index: []
(1, 2, 3)

piRSquared · Accepted Answer · 2018-08-31 16:02:08Z

2

You are attempting to subclass a dataframe.

I've answered how to subclass here -> LINK

Why have an attribute of a dataframe that is another dataframe? If you want an object that has several attributes, one of which is a dataframe, that's fine. Just don't define the class as class MyClass(pd.DataFrame): but class MyClass(object): instead.

answered Aug 31, 2018 at 16:02

piRSquared

296k68 gold badges509 silver badges654 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

Juergen Over a year ago

Thanks, I think that link will help a lot. I want an obejct, that behaves like a dataframe and has private members, some of which are again dataframes (or subclasses of dataframe). Is this conceptually bad style?

piRSquared Over a year ago

Probably. But I think you can accomplish your task if you pass the name of your attributes into the _metadata class variable.

Collectives™ on Stack Overflow

Python class derived from pandas DataFrame with list/DataFrame attribute

1 Answer 1

2 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related