Numpy average does not work on map object-CodePudding

I have an array that contains strings representing numbers.

a = ["101", "102", "103"]

I wanted to take the average of this array, so I tried mapping each element into a float and then using np.average

>>> np.average(map(float, a))
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "<__array_function__ internals>", line 5, in average
  File "C:\Users\alonso\AppData\Local\Programs\Python\Python39\lib\site-packages\numpy\lib\function_base.py", line 380, in average
    avg = a.mean(axis)
  File "C:\Users\alonso\AppData\Local\Programs\Python\Python39\lib\site-packages\numpy\core\_methods.py", line 190, in _mean
    ret = ret / rcount
TypeError: unsupported operand type(s) for /: 'map' and 'int'

If I convert the numbers by iteration it works, but if I have to use iteration this defeats the purpose of map.

np.average([float(i) for i in a])

102.0

Map returns a map object

map(float, a)
<map object at 0x0000022E46035D60>

But then, it looks like python doesn't know how to divide map by int.

If I convert map into a list, then it works, it doesn't create a list of maps, but rather evaluates the map value and creates a list of floats

list(map(float, a))
[101.0, 102.0, 103.0]

However converting map into an array doesn't.

np.array(map(float, a))
array(<map object at 0x0000022E2D4A84F0>, dtype=object)


np.average(np.array(map(float, a)))
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "<__array_function__ internals>", line 5, in average
  File "C:\Users\alonso\AppData\Local\Programs\Python\Python39\lib\site-packages\numpy\lib\function_base.py", line 380, in average
    avg = a.mean(axis)
  File "C:\Users\alonso\AppData\Local\Programs\Python\Python39\lib\site-packages\numpy\core\_methods.py", line 190, in _mean
    ret = ret / rcount
TypeError: unsupported operand type(s) for /: 'map' and 'int'
>>>

Why does this happen? Am I missing something or is an implementation limitation of numpy?

It looks counterintuitive to me that if I map something into floats I have to explicitly convert them into a list before taking the average, shouldn't np.average already handle that?

CodePudding user response：

Just forget about map. It provides poor readability and is rarely useful. List comprehensions and generator expressions have made it obsolete.

As for why it doesn't work: Map returns an iterator object in Python-3. It used to be a list in Python-2, which is why this might have worked in the past. However, numpy generally doesn't handle iterators gracefully. Use np.fromiter(map(float, a), dtype=float) or np.array(list(map(float, a)))

CodePudding user response：

In Python 3, map ceases to return a list object, so if you want a new list and not just something to iterate over, you either need list(map(float, mylist) or [float(i) for i in lst]