How to check if a variable is either a python list, numpy array or pandas series

Question

I have a function that takes in a variable that would work if it is any of the following three types

 1. pandas Series
 2. numpy array (ndarray)
 3. python list

Any other type should be rejected. What is the most efficient way to check this?

Solution 1

You can do it using isinstance:

import pandas as pd
import numpy as np
def f(l):
    if isinstance(l,(list,pd.core.series.Series,np.ndarray)):
        print(5)
    else:
        raise Exception('wrong type')

Then f([1,2,3]) prints 5 while f(3.34) raises an error.

Solution 2

Python type() should do the job here

l = [1,2]
s= pd.Series(l)
arr = np.array(l) 

When you print

type(l)
list

type(s)
pandas.core.series.Series

type(arr)
numpy.ndarray

Solution 3

The other answers are good but I sort of prefer this way:

if np.ndim(l)!=0:
    # this is something like a series, list, ndarray, etc.

It is nice because it provides more duck-typing flexibility compared to:

if isinstance(l, (pd.Series, list, np.ndarray)):
    # this is ONLY a pd.Series, list, or ndarray

...but it is better than this, which will allow a string or an iterator though-- both of which are often not wanted:

if isinstance(l, typing.Iterable):
    # this any iterable

...or this, which excludes a string but (weirdly) does not exclude an iterator:

if not np.isscalar(l):
    # this is something like a series, list, ndarray, etc.

However, if you truly only want a list, ndarray, or Series, the other answers are preferable.

Solution 4

This all really depends on what you are trying to achieve (will you allow a tuple, how about a range object?), but to be a bit less restrictive but still disallow strings (which I am guessing is what you are really trying to achieve) you can use the following code.

import collections
import pandas
import numpy

def myfunc(x):
    if not isinstance(x, collections.abc.Iterable) or isinstance(x, (str, bytes)):
        raise ValueError('A non-string iterable is required')
    return 'Yay!'

myfunc([9, 7])
myfunc((9, 7))
myfunc(numpy.arange(9))
myfunc(range(9))
myfunc(pandas.Series([9, 7]))
myfunc('Boo')  # THIS WILL RAISE A ValueError!!!!!

Solution 5

You can use isinstance like this:

import pandas as pd
import numpy as np

#Simple List
simple_list = [1,2]  

#numpy array
np_array = np.array(simple_list) 

#Pandas series
pandas_series = pd.Series(simple_list)  


if isinstance(simple_list, list):
    print("This is a list: ", simple_list)

if isinstance(np_array, np.ndarray):
    print("This is a numpy array: ", np_array)

if isinstance(pandas_series, pd.core.series.Series):
    print("This is pandas series: ", pandas_series)

View more solutions

Comments

[SethMMorton]

What about a tuple? I ask because you might be over-restricting your input.

[Zhang18]

So what do you suggest to check if the input is a list-like object, including np.ndarray and pd.Series? I was looking for something more elegant than the currently accepted answer anyway.

[Zhang18]

What's the difference between pd.core.series.Series and pd.Series?

[Miriam Farber]

@Zhang18, you could use pd.Series as well. I used pd.core.series.Series just because type(pd.Series([1,2,3])) gives pd.core.series.Series. Yet, both of them work, as pd.Series==pd.core.series.Series is True

[hBy2Py]

Likely, Series is actually defined in the pd.core.series module, but then the Series class is imported from there into the top-level __init__.py for convenience.