How to extract the index of a value in a numpy array?

Question

Here is the scenario:

I have the following variables:

val = [('am', '<f8'), ('fr', '<f8')] # val is type numpy.recarray     

am = [12.33, 1.22, 5.43, 15.23]  # am is type numpy.ndarray  

fr = [0.11, 1.23, 2.01, 1.01]   # fr is type numpy.ndarray

What I need is to detect the index for am = 12.33 and am = 15.23, once extracted (in this case indexes are [0] and [3]), I need to create the new variables:

new_am = [12.33, 15.23] 

new_fr = [0.11, 1.01]

My question is: Any idea about how to extract the indexes?

I have already used .index and np.where but it seems to have problems since I received an error message for .index:

"AttributeError: 'numpy.ndarray' object has no attribute ".index"

and for np.where the returning index is nothing array([], dtype=int64)

Thanks for any idea!

So you want the index as np.array([0, 3]) here? np.where(am == 12.33)[0] doesn't give you what you need? — akuiper
– akuiper, Commented Dec 28, 2016 at 2:06
Correct, and with that...produce the "new_am" and "new_fr" represented above. — FMEZA
– FMEZA, Commented Dec 28, 2016 at 2:08

akuiper · Accepted Answer · 2016-12-28 02:20:32Z

3

You may want the np.in1d which returns an array of boolean to indicate if an element is in another array:

import numpy as np
am = np.array([12.33, 1.22, 5.43, 15.23]) 
fr = np.array([0.11, 1.23, 2.01, 1.01])

index = np.where(np.in1d(am, [12.33, 15.23]))[0]
index
# array([0, 3])

am[index]
# array([ 12.33,  15.23])

fr[index]
# array([ 0.11,  1.01])

Or maybe you have an array with attributes:

new_arr = np.array(zip(am, fr), dtype=val)

index = np.where(np.in1d(new_arr['am'], [12.33, 15.23]))[0]

new_am = new_arr[index]['am']
new_fr = new_arr[index]['fr']

new_am
# array([ 12.33,  15.23])

new_fr
# array([ 0.11,  1.01])

edited Dec 28, 2016 at 2:20

answered Dec 28, 2016 at 2:12

akuiper

216k33 gold badges362 silver badges379 bronze badges

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

5 Comments

FMEZA Over a year ago

Thanks! interesting....seems to work for the values I gave you, but not for my real case values...all of them have exactly the same features of the values from the example above:

>>params['ampl'] ---> array([ 3931.41287385, 412.51629703, 257.47891338, ..., 193.89134605, 192.66807322, 192.20132863]) ---> index = np.where(np.in1d(params['ampl'], [3931.41287385]))[0] ---> index ---> array([], dtype=int64)

akuiper Over a year ago

That could be due to the float number inaccuracy. You can test this by params['ampl'][0] == 3931.41287385 to seem if they are equal. Some tolerance has to be introduced here.

FMEZA Over a year ago

Thanks...unfortunately same behavior index = np.where(params['ampl'][0] == 3931.41287385) ---> index ---> (array([], dtype=int64),)

akuiper Over a year ago

No, I mean literally print(params['ampl'][0] == 3931.41287385), you will see that it should be true but it returns false due to float number inaccuracy.

FMEZA Over a year ago

Thanks Psidom for your help! You were right, that was a inaccuracy in the float number, I used np.around to reduce the number to 3 decimals and worked perfect!

Collectives™ on Stack Overflow

How to extract the index of a value in a numpy array?

1 Answer 1

5 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

5 Comments

Your Answer

Sign up or log in

Post as a guest

Related