Reshape numpy array

Question

I have a numpy array X of shape (20000,3):

X = array([[  3,   8,   4],
           [  1,   2,   4],
           ...
           [  5,   8,   4],
           [  3,   9,   4]])

I want to drop the last 'column' (i.e. where the 4s are located), to get something like this:

X = array([[  3,   8],
           [  1,   2],
           ...
           [  5,   8],
           [  3,   9]])

Do I need to use a list comprehension to do this (i.e. iterate over array elements, save results to a list and convert list to array, or can I use a built-in numpy function, such as extract or where? Maybe you recommend something all together different? Thank you.

You can try this : y = X[:, :2]

Abid Rahman K
– Abid Rahman K

2017-07-10 06:18:11 +00:00
Commented Jul 10, 2017 at 6:18 — Abid Rahman K
– Abid Rahman K, Commented Jul 10, 2017 at 6:18

ksai · Accepted Answer · 2017-07-10 06:26:39Z

3

Use scipy or numpy, specify the array, column_to_delete, row/column

scipy.delete(X, 2, 1)   #(np.array, index, axis)
numpy.delete(X, 2, 1) # would give the same result

edited Jul 10, 2017 at 6:26

answered Jul 10, 2017 at 6:21

ksai

9776 silver badges18 bronze badges

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

Comments

JohanL · Accepted Answer · 2017-07-10 06:21:00Z

2

All you need to do is to pick out a slice of the original array:

X = np.array([[  3,   8,   4],
              [  1,   2,   4],
              [  5,   8,   4],
              [  3,   9,   4]])

newX = np.array(X[:,:-1])

Here I have chosen all rows and all but the last column of the original array. Then, in addition I have made it to a new array (the extra np.array() wrapping) in order to ensure it gets new memory. Otherwise it will just be a view of the original array and any updates to newX will also affect X.

answered Jul 10, 2017 at 6:21

JohanL

6,9211 gold badge16 silver badges32 bronze badges

2 Comments

Shai Over a year ago

why np.array for slicing X?

JohanL Over a year ago

@Shai To ensure a new array and thereby new memory layout. It is not strictly necessary, but as I wrote in the answer, it prevents updates to the new array not affect the original one.

Steve Geluso · Accepted Answer · 2017-07-10 06:21:58Z

1

import numpy
X = numpy.array([[  3,   8,   4],
           [  1,   2,   4],
           [  5,   8,   4],
           [  3,   9,   4]])
XX = X[:,:-1]
print(XX)

# [[3 8]
#  [1 2]
#  [5 8]
#  [3 9]]

answered Jul 10, 2017 at 6:21

Steve Geluso

1591 gold badge2 silver badges7 bronze badges

Comments

Achyuta nanda sahoo · Accepted Answer · 2017-07-10 06:27:42Z

1

you need to do slice op.

y = x[:,:2]

answered Jul 10, 2017 at 6:27

Achyuta nanda sahoo

4553 silver badges15 bronze badges

Collectives™ on Stack Overflow

Reshape numpy array

4 Answers 4

Comments

2 Comments

Comments

Comments

Hot Network Questions

Collectives™ on Stack Overflow

4 Answers 4

Comments

2 Comments

Comments

Comments

Related