remove rows by pick the min value in another column

Can you help?

On below matrix, for same X and Y, pick smallest Z, then put into a new matrix by removing the rows? and also give the row number for which row is kept.

X    Y   Z 
60,-40,1.1 
55,-40,3.1 
55,-45,4.5 
60,-45,5.3 
60,-40,7 
55,-40,8.4 
55,-45,9.8 
60,-40,11.2 
60,-45,12.6 
55,-40,14 
55,-45,15.4 
60,-45,16.8 
60,-40,18.2 
60,-45,19.6 
55,-40,21 
60,-40,22.4 
55,-45,23.8 
60,-45,25.2 
55,-40,26.6 
55,-45,28

[Image Analyst]

I can think of at least two ways. How many can you think of?

[Image Analyst]

And Jos thought of another the first time you asked this question. What was wrong with his answer?

[booterr]

can you also give the row number for which row is kept?

Output the row number (or row index)which is kept. For example, there is a total of 20 rows in the question. Finally, only row 1,2,3, and 4 is kept. Can you output this row number?

[Stephen23]

Duplicate question:

http://www.mathworks.com/matlabcentral/answers/279038-remove-rows-under-certain-condition

[booterr]

not a duplicate questions. For the first thread, the answer did not give which row number the Z is located in the initial matrix.

I am not only needing the value of Z, but also need the row number of which Z is outputted.

thanks

Answer 1

[Andrei Bobrov]

x=[55,-45,9.8 
60,-40,11.2 
60,-45,12.6 
55,-40,14 
55,-45,15.4 
60,-45,16.8 
60,-40,18.2 
60,-45,19.6 
55,-40,21 
60,-40,22.4 
55,-45,23.8 
60,-45,25.2 
55,-40,26.6 
55,-45,28]

[a,~,c] = unique(x(:,1:2),'rows');
n = size(a,1);
valuemin = zeros(n,4);
for ii = 1:n
    i0 = find(c == ii);
    [valuemin(ii,3),i1] = min(x(i0,end));
    valuemin(ii,[1:2,4]) = [a(ii,:),i0(i1)];
end

or

[a,~,c] = unique(x(:,1:2),'rows');
idx = accumarray(c,(1:size(x,1))',[],@(y)y(min(x(y,3))==x(y,3)));
valuemin = [a,x(idx,3),idx];