How to calculate each single element of a numpy array based on conditions

Question

I have a one column numpy array with multiple elements. I want to write a function that takes each element and evaluates if it meets a condition. Then depending on the result, it should use equations to calculate something. The calculation result should have the same size as the input array.

This is a simple example of what I want to do. (The actual code will be more complex). I know why it doesn't work but can't seem to find a solution.

import numpy as np
array1 = np.arange(1,11,1)

def test(array1):
    value1 = 20
    i = 0
    value3 = array1[i]
    while array1 > value3 and i < value1:
        i =+ 1
        value3 = array1[i]

test(array1)

I have tried to find a solution:

I looked at any() and all(), but it isnt't what I need, since I want to look at each element seperately.
I looked at numpy.where() but it return only the elements that meet a condition and therefor changes the array size, which isn't what I want.

EDIT: For my complete solution both the for loop version by Zulfiqaar and the where function as shown by Siva Kowuru are possible. Since the where function seems faster and more convenient I marked it as the accepted answer.

np.where is the way to go, see the answer of @SivaKowuru below. — FabienP
– FabienP, Commented Oct 21, 2017 at 12:23

FabienP · Accepted Answer · 2017-10-21 11:19:41Z

3

You could add the optional parameters on numpy.where(condition[, x, y]) to keep the same array size.

When True, yield x, otherwise yield y

https://docs.scipy.org/doc/numpy/reference/generated/numpy.where.html

To compare to the answer from @Zulfiqaar, the solution would be

np.where(array1 < 3, array1 ** 2, array1 / 3)

Note that the condition is just a bool array, so you can combine multiple conditions at once using binary operators. For instance comparing to another array:

array2 = np.random.randint(1, 11, 10)

>>> array2
Out[]: array([ 6,  1,  6,  2, 10,  4,  9, 10,  3,  3])

>>> np.where((array1 < 3) & (array1 > 5), array1 ** 2, array1 / 3)
Out[]:
array([ 0.33333333,  4.        ,  1.        ,  1.33333333,  1.66666667,
        2.        ,  2.33333333,  2.66666667,  3.        ,  3.33333333])

edited Oct 21, 2017 at 11:19

FabienP

3,1581 gold badge24 silver badges27 bronze badges

answered Oct 20, 2017 at 17:00

Siva Kovvuru

861 silver badge4 bronze badges

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

1 Comment

Weena Over a year ago

Thank you, for showing me what can be done with the np.where function!

Zulfiqaar · Accepted Answer · 2017-10-20 15:43:35Z

1

should use if for conditions:

import numpy as np

array1 = np.arange(1, 11, 1)
array2 = np.zeros(10)


for eachValue in range(len(array1)):                #iterate through array
    if (array1[eachValue] <3 ):                     #check for your condition
        array2[eachValue] = array1[eachValue] ** 2  #apply equation
    else:                                           #if condition not met
        array2[eachValue] = array1[eachValue] /  3  #do something else

edited Oct 20, 2017 at 15:43

answered Oct 20, 2017 at 15:37

Zulfiqaar

6431 gold badge6 silver badges12 bronze badges

6 Comments

Weena Over a year ago

Thank you that looks much better. What if I want to compare each value to another array to determine which one is closest to it and bigger? Can I use another loop inside this?

Zulfiqaar Over a year ago

you can use a loop within it - or you can do it without a second loop. Do you mean comparing each value to a specific value of another array, or its corresponding indexed value?

Weena Over a year ago

Yes, I want to find a specific value in another array, to determine the right pipe size. It needs to be bigger than the calculated ideal value stored in the first array. (So I have multiple pipes and need to find the appropriate pipe size.)

Zulfiqaar Over a year ago

Do the comparison inside the if statement: my condition was if value_in_first_array is less than 3, then square value - if it isnt (else) then divide by 3

Weena Over a year ago

Ok I will try that out. Thanks!

|

Spas · Accepted Answer · 2017-10-21 09:26:13Z

0

You can use map with lambda function instead of looping through all the elements:

value3 = array1[0]
map_array = np.array(list(map(lambda x: x**2 if x < value3 else x/3, array1)))

If you aren't using python 3 you won't need to convert it to list from the map function.

edited Oct 21, 2017 at 9:26

answered Oct 21, 2017 at 9:11

Spas

87016 silver badges14 bronze badges

Collectives™ on Stack Overflow

How to calculate each single element of a numpy array based on conditions

3 Answers 3

1 Comment

6 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

1 Comment

6 Comments

Comments

Your Answer

Sign up or log in

Post as a guest