How to change values in matrix to corresponding row value in vector based on condition

Question

I’m interested in whether it’s possible do the following with pure numpy.
Let’s say I have a matrix a and a vector b. I want to fill whatever values meet the condition to the left of the equal sign with the value from the vector b correponding to the row in matrix a.

import numpy as np
a = np.arange(30).reshape(5,6)
a[3,3] = 2
a[4,0] = 1
a
array([[ 0,  1,  2,  3,  4,  5],
       [ 6,  7,  8,  9, 10, 11],
       [12, 13, 14, 15, 16, 17],
       [18, 19, 20,  2, 22, 23],
       [ 1, 25, 26, 27, 28, 29]])

b = np.arange(0,5)[:,None]
b
array([[0],
       [1],
       [2],
       [3],
       [4]])

a[a<3] = b

TypeError: NumPy boolean array indexing assignment requires a 0 or 1-dimensional input, input has 2 dimensions

I understand why this doesn’t work and that I would need to reshape b to the same shape as a, and then subset it by a<3 first

b_mat = np.full(a.shape, b)
a[a<3] = b_mat[a<3]

a
array([[ 0,  0,  0,  3,  4,  5],
       [ 6,  7,  8,  9, 10, 11],
       [12, 13, 14, 15, 16, 17],
       [18, 19, 20,  3, 22, 23],
       [ 4, 25, 26, 27, 28, 29]])

However, I’d like to know, is there a way to do this elegantly in numpy without making the 1-dimensional vector b into a matrix where all the rows repeat?

There is a way to do this in Pandas as solved in this question.

Asked By: matsuo_basho

||

Source

Answer 1

Use a < 3 as mask and b as fill_value to get the initial array filled on masked values:

a = np.ma.array(a, mask=a < 3, fill_value=b).filled()
print(a)

[[ 0  0  0  3  4  5]
 [ 6  7  8  9 10 11]
 [12 13 14 15 16 17]
 [18 19 20  3 22 23]
 [ 4 25 26 27 28 29]]

Answered By: RomanPerekhrest

Answer 2

Why not use np.where?

>>> np.where(a < 3, b, a)
array([[ 0,  0,  0,  3,  4,  5],
       [ 6,  7,  8,  9, 10, 11],
       [12, 13, 14, 15, 16, 17],
       [18, 19, 20,  3, 22, 23],
       [ 4, 25, 26, 27, 28, 29]])

Answered By: Chrysophylaxs

How to change values in matrix to corresponding row value in vector based on condition

Question:

Answers: