How to modify rows of numpy arrays stored as a list

Question

I want to modify rows of numpy arrays stored in a list. length of my numpy arrays are not the same. I have several huge numpy arrays stored as list. This is my data (for simplicity I copied only a small list of array):

elements= [array([[971, 466, 697, 1, 15, 18, 28],
                 [5445, 4, 301, 2, 12, 47, 5]]),
           array([[5883, 316, 377, 2, 9, 87, 1]])]

Then, I want to replace the fourth column of each row with the last one and then delete the last column. I want to have the following result:

[array([[971, 466, 697, 1, 28, 18],
        [5445, 4, 301, 2, 5, 47]]),
 array([[5883, 316, 377, 2, 1, 87]])]

I tried the following code but it was not successful:

length=[len(i) for i in elements] # To find the length of each array
h=sum(length) # to find the total number of rows
for i in range (h):
    elements[:,[4,-1]] = elements[:,[-1,4]]
    elements=np.delete(elements,[-1],1)

I am facing the following error:

TypeError: list indices must be integers or slices, not tuple

I appreciate ay help in advance.

mathfux · Accepted Answer · 2020-10-26 11:17:26Z

2

You can do it without loops but it's still slower (1.75 times on large data) than accepted solution:

counts = list(map(len, elements))
arr = np.concatenate(elements)
arr[:, 4] = arr[:, -1]
new_elements = np.split(arr[:,:-1], np.cumsum(counts)[:-1])

Concatenation is quite slow in numpy.

answered Oct 26, 2020 at 11:17

mathfux

5,9792 gold badges20 silver badges38 bronze badges

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

Comments

Mathieu · Accepted Answer · 2020-10-26 10:11:06Z

1

A simple inefficient solution:

import numpy as np

elements= [np.array([[971, 466, 697, 1, 15, 18, 28],
                     [5445, 4, 301, 2, 12, 47, 5]]),
           np.array([[5883, 316, 377, 2, 9, 87, 1]])]

new_elements = list()
for arr in elements:
    arr[:, 4] = arr[:, -1]
    new_elements.append(arr[:, :-1])

The new list output is:

new_elements
Out[11]: 
[array([[ 971,  466,  697,    1,   28,   18],
        [5445,    4,  301,    2,    5,   47]]),
 array([[5883,  316,  377,    2,    1,   87]])]

answered Oct 26, 2020 at 10:11

Mathieu

5,8367 gold badges34 silver badges66 bronze badges

1 Comment

Link_tester Over a year ago

Dear @Mathieu, Thanks for your solution. It is exactly what I wanted. I do appreciate it.

user3103059 · Accepted Answer · 2020-10-26 10:47:43Z

1

Try this one

p=[]
for x in range(len(elements)):
    for y in range(len(elements[x])):
         p.append(list(elements[x][y][:4])+[elements[x][y][-1]]+[elements[x][y][-2]])
print(p)

[[971, 466, 697, 1, 28, 18],
[5445, 4, 301, 2, 5, 47],
[5883, 316, 377, 2, 1, 87]]

answered Oct 26, 2020 at 10:47

user3103059

Collectives™ on Stack Overflow

How to modify rows of numpy arrays stored as a list

3 Answers 3

Comments

1 Comment

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

Comments

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Related