What s the simplest way to extend a numpy array in 2 dimensions

Question

I have a 2d array that looks like this   XX xx   What s the most efficient way to add an extra row and column   xxy xxy yyy   For bonus points  I d like to also be able to knock out single rows and columns  so for example in the matrix below I d like to be able to knock out all of the a s leaving only the x s - specifically I m trying to delete the nth row and the nth column at the same time - and I want to be able to do this as quickly as possible   xxaxx xxaxx aaaaa xxaxx xxaxx

User · Accepted Answer

The shortest in terms of lines of code i can think of is for the first question.

>>> import numpy as np
>>> p = np.array([[1,2],[3,4]])

>>> p = np.append(p, [[5,6]], 0)
>>> p = np.append(p, [[7],[8],[9]],1)

>>> p
array([[1, 2, 7],
   [3, 4, 8],
   [5, 6, 9]])

And the for the second question

    p = np.array(range(20))
>>> p.shape = (4,5)
>>> p
array([[ 0,  1,  2,  3,  4],
       [ 5,  6,  7,  8,  9],
       [10, 11, 12, 13, 14],
       [15, 16, 17, 18, 19]])
>>> n = 2
>>> p = np.append(p[:n],p[n+1:],0)
>>> p = np.append(p[...,:n],p[...,n+1:],1)
>>> p
array([[ 0,  1,  3,  4],
       [ 5,  6,  8,  9],
       [15, 16, 18, 19]])

User · Answer

You can use     gt  gt  gt  np concatenate  array1  array2           e g    gt  gt  gt  import numpy as np  gt  gt  gt  a     1  2  3   10  20  30    gt  gt  gt  b     100 200 300    gt  gt  gt  a   np array a    not necessary  but numpy objects prefered to built-in  gt  gt  gt  b   np array b        gt  gt  gt  a array    1   2   3           10  20  30     gt  gt  gt  b array   100  200  300     gt  gt  gt  c   np concatenate  a b    gt  gt  gt  c array     1    2    3            10   20   30           100  200  300     gt  gt  gt  print c     1   2   3     10  20  30    100 200 300      - - - - - -    Sometimes  you will come across trouble if a numpy array object is initialized with incomplete values for its shape property  This problem is fixed by assigning to the shape property the tuple   array length  element length    Note  Here   array length  and  element length  are integer parameters  which you substitute values in for  A  tuple  is just a pair of numbers in parentheses   e g    gt  gt  gt  import numpy as np  gt  gt  gt  a   np array   1 2 3   10 20 30     gt  gt  gt  b   np array  100 200 300     initialize b with incorrect dimensions  gt  gt  gt  a shape  2  3   gt  gt  gt  b shape  3    gt  gt  gt  c   np concatenate  a b    Traceback  most recent call last     File   lt pyshell 191 gt    line 1  in  lt module gt      c   np concatenate  a b   ValueError  all the input arrays must have same number of dimensions  gt  gt  gt  b shape    1 3   gt  gt  gt  c   np concatenate  a b    gt  gt  gt  c array     1    2    3            10   20   30           100  200  300

User · Answer

A useful alternative answer to the first question  using the examples from tomeedee   s answer  would be to use numpy   s vstack and column stack methods   Given a matrix p    gt  gt  gt  import numpy as np  gt  gt  gt  p   np array    1 2     3 4       an augmented matrix can be generated by    gt  gt  gt  p   np vstack    p    5   6       gt  gt  gt  p   np column stack    p     7   8   9        gt  gt  gt  p array   1  2  7           3  4  8           5  6  9      These methods may be convenient in practice than np append   as they allow 1D arrays to be appended to a matrix without any modification  in contrast to the following scenario    gt  gt  gt  p   np array     1   2       3   4       5   6        gt  gt  gt  p   np append  p     7   8   9     1   Traceback  most recent call last     File   lt stdin gt    line 1  in  lt module gt    File   usr lib python2 6 dist-packages numpy lib function base py   line 3234  in append     return concatenate  arr  values   axis axis  ValueError  arrays must have same number of dimensions   In answer to the second question  a nice way to remove rows and columns is to use logical array indexing as follows   Given a matrix p    gt  gt  gt  p   np arange  20   reshape    4   5       suppose we want to remove row 1 and column 2    gt  gt  gt  r   c   1   2  gt  gt  gt  p   p   np arange  p shape 0       r         gt  gt  gt  p   p       np arange  p shape 1       c    gt  gt  gt  p array    0   1   3   4           10  11  13  14           15  16  18  19      Note - for reformed Matlab users - if you wanted to do these in a one-liner you need to index twice    gt  gt  gt  p   np arange  20   reshape    4   5          gt  gt  gt  p   p   np arange  p shape 0       r             np arange  p shape 1       c     This technique can also be extended to remove sets of rows and columns  so if we wanted to remove rows 0  amp  2 and columns 1  2  amp  3 we could use numpy s setdiff1d function to generate the desired logical index    gt  gt  gt  p   np arange  20   reshape    4   5      gt  gt  gt  r     0   2    gt  gt  gt  c     1   2   3    gt  gt  gt  p   p   np setdiff1d  np arange  p shape 0     r           gt  gt  gt  p   p       np setdiff1d  np arange  p shape 1      c      gt  gt  gt  p array    5   9           15  19

User · Answer

Answer to the first question   Use numpy append   http   docs scipy org doc numpy reference generated numpy append html numpy append  Answer to the second question   Use numpy delete  http   docs scipy org doc numpy reference generated numpy delete html

User · Answer

I find it much easier to  extend  via assigning in a bigger matrix  E g   import numpy as np p   np array   1 2    3 4    g   np array range 20   g shape    4 5  g 0 2  0 2    p   Here are the arrays   p     array   1  2           3  4      g   array    0   1   2   3   4            5   6   7   8   9           10  11  12  13  14           15  16  17  18  19      and the resulting g after assignment      array    1   2   2   3   4            3   4   7   8   9           10  11  12  13  14           15  16  17  18  19

User · Answer

Another elegant solution to the first question may be the insert command   p   np array   1 2   3 4    p   np insert p  2  values 0  axis 1    insert values before column 2   Leads to   array   1  2  0           3  4  0      insert may be slower than append but allows you to fill the whole row column with one value easily   As for the second question  delete has been suggested before   p   np delete p  2  axis 1    Which restores the original array again   array   1  2           3  4

User · Answer

maybe you need this    gt  gt  gt  x   np array  11 22    gt  gt  gt  y   np array  18 7 6    gt  gt  gt  z   np array  1 3 5    gt  gt  gt  np concatenate  x y z   array  11  22  18   7   6   1   3   5

[python] What's the simplest way to extend a numpy array in 2 dimensions?

Examples related to python

Examples related to arrays

Examples related to math

Examples related to numpy