Replace number in specific indexes of a numpy array to NaN

Question

I have a numpy array that looks like this:

[[1399 17 4 3 0 0 0 0]
 [ 11 374 2 3 1 4 0 1]
 [ 7 0 187 4 0 0 1 1]
 [ 2 3 4 308 0 0 0 3]
 [ 2 0 0 0 280 3 0 1]
 [ 0 2 0 0 2 81 0 3]
 [ 1 0 2 0 2 0 154 4]
 [ 0 0 1 2 1 1 8 552]]

I want to replace the values where [x,x] that is the column and row number are the same, to np.nan. I have tried doing this:

for i in range(8):
 data[i][i] = np.nan

but it gives an error

Many thanks

Answer 1

Use numpy.fill_diagonal after typing the array into float:

import numpy as np
arr = np.random.randint(0, 10, (10,10))
arr = arr.astype(float)
np.fill_diagonal(arr, np.nan)

Output:

array([[nan, 2., 6., 6., 1., 2., 7., 2., 7., 0.],
 [ 3., nan, 3., 7., 3., 5., 2., 9., 1., 3.],
 [ 4., 2., nan, 2., 6., 0., 4., 3., 5., 8.],
 [ 8., 8., 6., nan, 7., 8., 4., 1., 0., 8.],
 [ 5., 2., 1., 5., nan, 6., 4., 7., 8., 7.],
 [ 9., 5., 7., 1., 6., nan, 7., 9., 2., 8.],
 [ 5., 1., 5., 6., 1., 0., nan, 5., 1., 6.],
 [ 5., 9., 7., 4., 4., 4., 0., nan, 3., 5.],
 [ 8., 2., 2., 0., 4., 5., 2., 2., nan, 3.],
 [ 7., 4., 9., 4., 5., 4., 2., 4., 0., nan]])

• Huzo: Thanks. One question, how would I later on remove the decimal points, and turn each value back to integer? doing the same conversion to integer seems to be wrong, that is if I do data = data.astype(int) afterwards
• Aly Hosny: nan recognizer as a float. you can't convert it to integer.