Indexing highest value of numpy matrix

Question

I have a numpy array of shape (4, 7) like this:

array([[ 1,  4,  5,  7,  8,  6,  7]
       [ 2, 23,  2,  4,  8, 94,  2],
       [ 1,  5,  6,  7, 10, 15, 20],
       [ 3,  9,  2,  7,  6,  5,  4]])

I would like to get the index of the highest element, i.e. 94, in a form like: first row fifth column. Thus the output should be a numpy array ([1,5]) (matlab-style).

Answer 1

You get the index of the maximum index using arr.argmax() but to get the actual row and column you must use np.unravel_index as below:

import numpy as np

arr = np.array([[ 1,  4,  5,  7,  8,  6,  7],
                [ 2, 23,  2,  4,  8, 94,  2],
                [ 1,  5,  6,  7, 10, 15, 20],
                [ 3,  9,  2,  7,  6,  5,  4]])

maximum = np.unravel_index(arr.argmax(), arr.shape)

print(maximum)
# (1, 5)

You have to use np.unravel_index as by default np.argmax will return the index from a flattened array (which in your case would be index 12).