如何从数字列表中创建数字符号的列表/数组

Question

I have a numpy array of numbers:

n =  [ 1.2,0,-0.5,0.3,0,-0.8]

I want to create a numpy array using the above that only holds the sign of the numbers, result should be:

s = [1,0,-1,1,0,-1]

I can create this with a loop:

s= np.zeros(n.shape[0])    
for i in range (n.shape[0]):
    if n[i]>0: s[i]=1
    if n[i]<0: s[i]=-1    

Is there a way to use list comprehension with numpy arrays that can do the same with high performance?

Answer 1

If you are using numpy, a better solution is to use numpy.sign():

import numpy as np
s = np.sign(n)

This will give you a numpy array.

array([ 1., 0., -1., 1., 0., -1.])

To convert this floating point result to int, you may use:

s.astype(np.int)

If you want to convert it back to python list:

s_list = s.tolist()

You can do the above in one line as:

s = np.sign(n).astype(np.int).tolist()

Answer 2

The np.sign answer seems the best way, but if you want to code something, I feel like this should be pretty quick:

import numpy as np

def get_signs(array_of_numbers):
    f = lambda x: x and (1, -1)[x < 0]
    return np.fromiter((f(item) for item in array_of_numbers), array_of_numbers.dtype)