[Python] What is the good practice to add many similar methods to a class in Python?

Let's say I have a class like this.

from scipy import interpolate
import numpy as np

class Bar:
def __init__(self,data):
self.data = data

def mean(self):
return self.data.mean(axis=0)

def sd(self):
return self.data.std(axis=0)

bar = Bar(np.random.rand(10,5))
print(bar.mean())
print(bar.sd())


The class Bar may have many such methods such as mean(), sd() etc. I want to add sampled versions of those methods, so that I can simply get results equivalent to this:

new_ids = np.linspace(bar.ids[0],bar.ids[-1],100)
sampled_mean = interpolate.interp1d(bar.ids,bar.mean(),axis=0)(new_ids)


Current workaround: manually add new methods with help of decorators.

from scipy import interpolate
import numpy as np

def sample(func):
def wrapper(self,n_sample=100,*args,**kwargs):
new_ids = np.linspace(self.ids[0],self.ids[-1],n_sample)
vals = func(self,*args,**kwargs)
return interpolate.interp1d(self.ids,vals,axis=0)(new_ids)
return wrapper

class Bar:
def __init__(self,ids,data):
self.ids = ids
self.data = data

def mean(self):
return self.data.mean(axis=0)

def sd(self):
return self.data.std(axis=0)

@sample
def spl_mean(self):
return self.mean()

@sample
def spl_sd(self):
return self.sd()

bar = Bar(ids=np.arange(5),data=np.random.rand(10,5))

new_ids = np.linspace(bar.ids[0],bar.ids[-1],100)
sampled_mean = interpolate.interp1d(bar.ids,bar.mean(),axis=0)(new_ids)
assert np.all(sampled_mean == bar.spl_mean())


However, I have many such methods. Of course I can write them with the help of LLM but what I want to know is:

• whether this is already a good practice to achieve this, and
• whether there's more elegant approache.

Continue reading...