CODEWITHSUNDEEP
pythonsortingnumpymatrix
Sam H.
Sam H.

Reputation: 33

Append/Concatenate & sort two Numpy matrixes

I have one input array with a shape of [1,500] in the form:

[[0][1][2] ... [499]]

To be combined with a random output array with a shape of [1,500] as:

[[22][16][11] ... [51]]

I have attempted using either concatenate or append like so (1 at a time):

inAndout = np.append(soloInput, outputMatrix, axis=1) #or:
inAndout = np.append(soloInput, outputMatrix, axis=1) #Commented 1 out each time

The output from both is fairly good with a form when print(inAndOut):

[[0 22][1 16][2 11] ... [499 51]] #Notice no ',' only spacing between added arrays

When I then use:

sortedInput = np.sort(inAndout, axis=0)

It results in:

[[0 11][1 16][2 22] ... [499 51]] #It sorts both inputs and outputs

Also when I use axis=1 it results in an unsorted matrix.

What I would like is for it to just sort based off outputs or just sort based off inputs like so:

[[0 22][1 16][2 22] ... [499 51]] #Sorted by inputs [[2 11][1 16][0 22] ... [499 51]] #Sorted by outputs

Any help is greatly appreciated!

Upvotes: 0

Views: 310

Answers (1)

akuiper
akuiper

Reputation: 214977

You can use argsort to return the indices that sorts either the input or the output and then reorder all columns with the indices:

Sort by input:

inAndout[inAndout[:,0].argsort(),:]

Sort by output:

inAndout[inAndout[:,1].argsort(),:]

Example:

inAndout = np.array([[0, 22],[1, 16],[2, 11],[499, 51]])

inAndout

#array([[  0,  22],
#       [  1,  16],
#       [  2,  11],
#       [499,  51]])

inAndout[inAndout[:,0].argsort(),:]

#array([[  0,  22],
#       [  1,  16],
#       [  2,  11],
#       [499,  51]])

inAndout[inAndout[:,1].argsort(),:]

#array([[  2,  11],
#       [  1,  16],
#       [  0,  22],
#       [499,  51]])

Upvotes: 1

Related Questions