Applying function to subclass of Pandas DataFrame just returns DataFrame and now Subclass

Question

I am trying to subclass Pandas' DataFrame object.

class AbundanceFrame(pd.DataFrame):
   'Subclass of DataFrame used for creating simulated dataset with various component timeseries'

    def __init__(self, days,*args,**kw):
        'Constructor for AbundanceFrame class, must pass index of timeseries'
        super(AbundanceFrame,self).__init__(index = days,*args,**kw)
        self.steps = 0
        self.monotonic = 0

I have a number of other methods that add simulated time-series to the resulting AbundanceFrame. The resulting Abundance frame takes on this form:

I then want to apply poisson sampling noise to all data in the abundance frame.

def apply_poisson_noise(self,keys=False):
    temp = self.copy()
    #print type(temp)
    if keys != False: 
        for key in keys:
            temp[key] = np.random.poisson(self[key])            
    else: 
        temp = self.apply(np.random.poisson)
    return temp

With the above I can create an AbundanceFrame without problems. However when I attempt to apply_poisson_noise() it returns a DataFrame and not an AbundanceFrame. I have been searching online and have not found a method for applying functions to DataFrames in place for pandas.

I would like to know how I can have this functionality and return an AbundanceFrame.

Thank you!

Answer 1

Solved the problem: (Building on the response of user4589964) In apply_poisson_noise() I just call the AbundanceFrame constructor and give it the calculated data.

from copy import deepcopy

class AbundanceFrame(pd.DataFrame):
'Subclass of DataFrame used for creating simulated dataset with various component timeseries'

def __init__(self, days,steps=0,monotonic=0,*args,**kw):
    'Constructor for AbundanceFrame class, must pass index of timeseries'
    super(AbundanceFrame,self).__init__(index = days,*args,**kw)
    self.steps = steps
    self.monotonic = monotonic

def apply_poisson_noise(self,keys=False):
    temp = deepcopy(self)
    if keys != False: 
        for key in keys:
            temp[key] = np.random.poisson(self[key])  
        temp =  AbundanceFrame(self.index, columns=self.columns, data = temp.values,
                               steps=self.steps, monotonic=self.monotonic)
    else: 
        temp =  AbundanceFrame(self.index, columns=self.columns, data = self.apply(np.random.poisson),
                               steps=self.steps, monotonic=self.monotonic)
    return temp