Create new array while Iterating over a n dimensional numpy array

Question

How do I iterate over a numpy array of n dimensions and create a new array of similar shape.

e.g. for the inputs:

import numpy as np

arr = np.array([[1,2,3],
                [4,5,6],
                [7,8,9],
                [0,0,0]])
alpha = 3.
median = np.median(arr)

I would like to build a new array of same (4,3) with flags set to 1 for a random condition. e.g.

flag = (arr[i,j] > median - alpha) or (arr[i,j] < median + alpha)

I would solve this with 2 for statements

flags = arr * 0 
for i in range(arr.shape[0]):
    for j in range(arr.shape[1]):
         flags[i,j] = (arr[i,j] > median - alpha) or (arr[i,j] < median + alpha)

Is there a way to solve this in a simpler and more efficient pythonic way ? The solution should ideally also work for n dimensional arrays (1,2, ... n dimensions)

Umm, right now you're generating True flags unless arr[i,j] == median * alpha. Is that what you want? — Daniel F
– Daniel F, Commented Oct 17, 2017 at 9:32
True, this is though just an example. What I am asking is the iteration process. I will update the condition to make a bit more sense — Valerian
– Valerian, Commented Oct 17, 2017 at 9:35

Daniel F · Accepted Answer · 2017-10-17 09:42:29Z

5

You don't have to iterate at all.

np.logical_or(arr < median - alpha, arr > median + alpha)

answered Oct 17, 2017 at 9:42

Daniel F

14.5k2 gold badges34 silver badges59 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

B. M. · Accepted Answer · 2017-10-18 12:06:01Z

1

Numpy is done to avoid loop:

alpha = 3.
median = np.median(arr)

abs(arr-median) < alpha

#array([[ True,  True,  True],
#      [ True,  True,  True],
#      [False, False, False],
#      [False, False, False]], dtype=bool)

And of course it works for any number of dimensions.

More generally, you can avoid loops each time you can formulate your condition with numpy logic functions. Python and and or operator do not map on numpy arrays, and using them implies loop and slow down.

edited Oct 18, 2017 at 12:06

answered Oct 17, 2017 at 13:01

B. M.

18.7k2 gold badges40 silver badges56 bronze badges

1 Comment

Valerian Over a year ago

Sure but your answer is "using" the particular condition that I set. In some cases though this condition may be very complex (or even worse may not be a condition at all) so I was wondering if there is a a more generic solution to avoid these kind of loops.

Collectives™ on Stack Overflow

Create new array while Iterating over a n dimensional numpy array

2 Answers 2

Comments

1 Comment

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

1 Comment

Related