select rows based on the value of the last column of the matrix

I have a matrix 'A'. As can be seen in the figure (red rectangles), some rows have the first and second columns the same but not the third column. Considering one of the red rectangles, I would like to select the entire row (green rectangle) that has a greater value in the fourth column. The end result is the matrix 'A_out'.

A = [334	1358	175	3
335	1359	176	3
335	1359	175	1
336	1360	176	2
337	1361	176	6
337	1361	177	1
338	1362	177	5
339	1363	177	10
340	1364	178	1
340	1364	177	10];

A_out = [334	1358	175
335	1359	176
336	1360	176
337	1361	176
338	1362	177
339	1363	177
340	1364	177];

Answer 1

[Matt J]

This solution requires only 1 sorting operation.

A = [334	1358	175	3
335	1359	176	3
335	1359	175	1
336	1360	176	2
337	1361	176	6
337	1361	177	1
338	1362	177	5
339	1363	177	10
340	1364	178	1
340	1364	177	10];

B=sortrows(A,[1,2,4]);

I=diff([findgroups(B(:,1), B(:,2));0])~=0;

A_out=B(I,1:3)

A_out = 7×3

334 1358 175 335 1359 176 336 1360 176 337 1361 176 338 1362 177 339 1363 177 340 1364 177

[Jon]

This is a neat approach. I haven't used splitapply very much, but I can see here how powerful it is.

Answer 2

[Jon]

A = [334	1358	175	3
335	1359	176	3
335	1359	175	1
336	1360	176	2
337	1361	176	6
337	1361	177	1
338	1362	177	5
339	1363	177	10
340	1364	178	1
340	1364	177	10];

% Sort by first and then third column
% use descending order so that when first column values are the same
% rows with bigger 3rd column values
% will occur first
Asrt = sortrows(A,[1,3],"descend")

Asrt = 10×4

340 1364 178 1 340 1364 177 10 339 1363 177 10 338 1362 177 5 337 1361 177 1 337 1361 176 6 336 1360 176 2 335 1359 176 3 335 1359 175 1 334 1358 175 3

% Get location of first instance of each unique first column value (by default sorted
% in increasing order), that is location for 334,335,336...
[~,ia] = unique(Asrt(:,1))

ia = 7×1

10 8 7 5 4 3 1

% Choose rows with largest 3rd column value
A_out = Asrt(ia,(1:3))

A_out = 7×3

334 1358 175 335 1359 176 336 1360 176 337 1361 177 338 1362 177 339 1363 177 340 1364 178

[Voss]

I like this answer, but I think to get the OP's intent ("first and second columns the same ... select the entire row that has a greater value in the fourth column"), a couple of modifications are needed:

A = [334	1358	175	3
335	1359	176	3
335	1359	175	1
336	1360	176	2
337	1361	176	6
337	1361	177	1
338	1362	177	5
339	1363	177	10
340	1364	178	1
340	1364	177	10];

Asrt = sortrows(A,[1 2 4],"descend");  % sort by column 1, then 2, then 4
[~,ia] = unique(Asrt(:,[1 2]),'rows'); % unique according to first two columns
A_out = Asrt(ia,1:3)

A_out = 7×3

334 1358 175 335 1359 176 336 1360 176 337 1361 176 338 1362 177 339 1363 177 340 1364 177

[Jon]

@Voss, Oh, I see now that in the OP's written description they say to select the "entire row", but in the example output A_out, they only show the first three columns. I was just looking at the example. Anyhow the OP can now see how to do it either way.

Oops, you're right my code selected the one with the larger value in the third column, I didn't read the problem description carefully enough.

Also, good catch on the first two columns being unique rather than just the first as I had. For this example I don't think it makes any difference, but your correction follows the OP's specification, thanks.