12

I have a Numpy ndarray matrix of float values and I need to select spesific rows where certain columns have values satisfying certain criteria. For example lets say I have the following numpy matrix: 

matrix = np.ndarray([4, 5])
matrix[0,:] = range(1,6)
matrix[1,:] = range(6,11)
matrix[2,:] = range(11,16)
matrix[3,:] = range(16,21)

Lets say I want to select rows from the matrix where the first column's value is between 1 and 6 and the value of second column is between 2-7.

How can I get the row-indexes of the matrix where these conditions are satisfied? What about if I want to delete the rows that satisfy the conditional criterion?

Improve this question

4 Answers 4

Reset to default
21

For a numpy based solution, you can use numpy.where and then get the row indexes from it and then use it for indexing you matrix. Example -

matrix[np.where((1 <= matrix[:,0]) & (matrix[:,0] <= 6)
       & (2 <= matrix[:,1]) & (matrix[:,1] <= 7))]

Demo -

In [169]: matrix
Out[169]:
array([[  1.,   2.,   3.,   4.,   5.],
       [  6.,   7.,   8.,   9.,  10.],
       [ 11.,  12.,  13.,  14.,  15.],
       [ 16.,  17.,  18.,  19.,  20.]])

In [170]: matrix[np.where((1 <= matrix[:,0]) & (matrix[:,0] <= 6)
   .....:        & (2 <= matrix[:,1]) & (matrix[:,1] <= 7))]
Out[170]:
array([[  1.,   2.,   3.,   4.,   5.],
       [  6.,   7.,   8.,   9.,  10.]])

Another method , as indicated in the comments would be to use boolean masks, Example -

mask = ((1 <= matrix[:,0]) & (matrix[:,0] <= 6)
           & (2 <= matrix[:,1]) & (matrix[:,1] <= 7))

matrix[mask,:]

Demo -

In [41]: matrix
Out[41]:
array([[  1.,   2.,   3.,   4.,   5.],
       [  6.,   7.,   8.,   9.,  10.],
       [ 11.,  12.,  13.,  14.,  15.],
       [ 16.,  17.,  18.,  19.,  20.]])

In [42]: mask = ((1 <= matrix[:,0]) & (matrix[:,0] <= 6)
   ....:            & (2 <= matrix[:,1]) & (matrix[:,1] <= 7))

In [43]:

In [43]: matrix[mask,:]
Out[43]:
array([[  1.,   2.,   3.,   4.,   5.],
       [  6.,   7.,   8.,   9.,  10.]])
Improve this answer
Sign up to request clarification or add additional context in comments.

3 Comments

Thank you very much! I'm used to more into Matlab's indexing so Python's indexing sometimes confuses me ;D
You could omit where and just use the boolean mask to select the rows. mask=(1<=matrix...); matrix[mask,:].
@hpaulj Good suggestion, thank you. I have included that in the answer.
1

You can get the indices with :

rows = np.logical_and(0 < matrix[:, 0], < matrix[:, 0] < 6 ) *  np.logical_and(1 < matrix[:, 1], matrix[:, 1] < 7)

Then newMatrix = np.delete(matrix, rows, axis = 0)

Improve this answer

1 Comment

Thank you, excellent! P.S. did you forget to add the 'rows' to the second line?
1

You mentioned MATLAB. Here's the equivalent to the accepted answer using Octave

octave:17> ma=reshape(1:20,5,4)
ma =
    1    6   11   16
    2    7   12   17
    3    8   13   18
    4    9   14   19
    5   10   15   20

octave:18> mask=(1<=ma(1,:))&(ma(1,:)<=6)&(2<=ma(2,:))&(ma(2,:)<=7)
mask =
   1   1   0   0

octave:19> ma(:,mask)
ans =
    1    6
    2    7
    3    8
    4    9
    5   10

The accepted answer without where is:

In [592]: mask=(1 <= matrix[:,0]) & (matrix[:,0] <= 6) &(2 <= matrix[:,1]) & (matrix[:,1] <= 7)

In [593]: matrix[mask,:]
Out[593]: 
array([[  1.,   2.,   3.,   4.,   5.],
       [  6.,   7.,   8.,   9.,  10.]])

I switched rows and columns in the Octave version because that is its natural way of generating the same numbers (MATLAB/Octave use the equivalent of numpys 'F' order - see below).

The other changes are 0 v 1 start index, and () v []. Otherwise the two notations are similar.

A simpler way to generate the matrix in numpy:

In [594]: np.arange(1,21).reshape(4,5)
Out[594]: 
array([[ 1,  2,  3,  4,  5],
       [ 6,  7,  8,  9, 10],
       [11, 12, 13, 14, 15],
       [16, 17, 18, 19, 20]])

Or with the MATLAB layout:

In [595]: np.arange(1,21).reshape(5,4,order='F')
Out[595]: 
array([[ 1,  6, 11, 16],
       [ 2,  7, 12, 17],
       [ 3,  8, 13, 18],
       [ 4,  9, 14, 19],
       [ 5, 10, 15, 20]])
Improve this answer

1 Comment

Thank you very much! =) Excellent!
0

Get row indices:

row_indices = [x for x in range(4) if matrix[x][0] in range(1,7) and matrix[x][1] in range(2,8)]

Delete rows:

indices = [x for x in range(4) if not( matrix[x][0] in range(1,7) and matrix[x][1] in range(2,8))]

new_matrix = matrix[indices]
Improve this answer

3 Comments

Thank you very much! =) Appreciate it
The part "in range(1, 7)" looks terribly inefficient
There might be even a more elegant method using numpy.where() ...still thinking about it.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.