CODEWITHSUNDEEP
pythonpython-3.xpandasnumpy
user8864088
user8864088

Reputation:

vectorized way to change numpy array values based on another array

Is there is a vectorized (or better) way of setting values to certain data points of numpy array based on another way other than this way?

import numpy as np

data = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
pos = np.array([[1, 2], [2, 0]])

for p in pos:
    i,j = p
    data[i,j] = 20

print(data)

Upvotes: 2

Views: 336

Answers (2)

user8864088
user8864088

Reputation:

Just extending the great answer by @piRSquared.

import numpy as np

data = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9],[10,11,12]])
pos = np.array([[1.0, 2.0,0.03], [2.0, 0.0,0.06]])

data = data.astype(float)

data[pos[:,0:2].astype(int).tolist()] = pos[:,-1]
data

array([[  1.  ,   2.  ,   3.  ],
       [  4.  ,   5.  ,   0.03],
       [  0.06,   8.  ,   9.  ],
       [ 10.  ,  11.  ,  12.  ]])

Upvotes: 0

piRSquared
piRSquared

Reputation: 294258

With later versions of Python you can create a list within a comprehension by unpacking another iterable. We then pass that list to do slice assignment.

The way we are accessing (slicing) is done via Integer Array Indexing

data[[*pos]] = 20
data

array([[ 1,  2,  3],
       [ 4,  5, 20],
       [20,  8,  9]])

For other versions of Python try:

data[pos.tolist()] = 20
data

Upvotes: 3

Related Questions