方法和属性python之间的区别？

Question

I'm probably using the wrong terminology here, but I'm looking for an explanation on the difference between a function called on an object, ie method, and then an attribute of the object itself.

Take the following df:

df = pd.DataFrame({'a':[4, 5, 6, 7],
                   'b':[3, 2, 5, 4]})

You can call the sum() method/function on df, and you can call the columns attribute on df. I may be getting confused between chaining a method vs. calling an attribute using . .

df.sum()
df.columns

Answer 1

They're the same. Methods are pretty much just attributes that happen to be of type function or any other runnable type. If you do

>>> f = df.sum

it'll work perfectly fine. If you then check the type, you'll see

>>> type(f)
<type 'instancemethod'>

or something similar along those lines. You can call any object of that type as a function, which is why you can do df.sum() : first it's finding the attribute of df that happens to be a function, and then it's calling it.

Answer 2

You can't "call" the df.columns part of a DataFrame. It's a data part of a DataFrame, and you "reference" it. df.sum() you can "call", because sum() is an executable part of the DataFrame.

The tricky bit (and I'm trying to avoid ambiguous language here) is that a DataFrame has both of those kinds of things: data parts and code parts. They can both change, be inherited, etc. But they're really different: You can't "call" columns, and you (generally) don't reference sum() .