Eliminating Multiple rows using 'if' conditional

Is it possible to eliminate multipe rows of a matrix using IF conditional?

 A=[1  1875  507
1  1880  508 %repeat 
1  1885  508 %repeat (eliminate this row)
1  1890  508 %repeat 
1  1895  512
2  1720  501
2  1730  502 %repeat 
2  1740  502 %repeat (eliminate this row)
2  1750  502 %repeat (eliminate this row)
2  1760  502 %repeat 
2  1770  505]

 Here, A(3,:), A(8:9,:) are to be eliminated. 
So that the the result would look like:
 B=[1  1875  507
1  1880  508
1  1890  508
1  1895  512
2  1720  501
2  1730  502
2  1760  502
2  1770  505]

How can such elimination be done using IF loop for a matrix bigger than this where it is very inefficient to do so for every unique value in Column1?

Thank you

Answer 1

[the cyclist]

Given your very special requirement for first and last rows, you might be able to use the unique() command. Does this work for you?

[~,indexToFirst] = unique(A(:,[1 3]),'rows','first');
[~,indexToLast]  = unique(A(:,[1 3]),'rows','last');

indexToFirstAndLastRows = union(indexToFirst,indexToLast);

A = A(indexToFirstAndLastRows,:)

[Walter Roberson]

Oh, good idea!

[MountainKing]

It works as desired! Thank you!!

Answer 2

[Walter Roberson]

Yes, it is possible to eliminate multiple rows using "if" conditions. For example,

if 1 + 1 == 2
  A([3 8:9], :) = [];
end

Often it is cleaner and more efficient to use logical indexing rather than "if" conditions.

The pattern you are looking for seems to be "within each unique value for the first column, if there are more than two consecutive rows with the same value for the third column, then eliminate all except the first and the last of them." Is that what you are looking for?

If so then what should happen if we add an additional entry to the end of the matrix,

2 1725 502

then should there be an implicit re-sorting to move the 1725 to between the 1720 and 1730, thus triggering the 1730 entry to be eliminated? Or since the 1725 entry would be "after" the other entries, should the 1740, 1750, 1760 and 1770 entries all be removed, even though the 1770 is a different column 3, under the rule that "all" entries between the first and last with the same third value should be eliminated?

[MountainKing]

Yes, the pattern we are looking for is exactly the one you described: "within each unique value for the first column, if there are more than two consecutive rows with the same value for the third column, then eliminate all except the first and the last of them."

And NO, for all 2s in row1, ONLY the rows below can be deleted:

 2  1740  502 %repeat (eliminate this row)
2  1750  502 %repeat (eliminate this row)

CANNOT remove the rows with 1760 & 1770 in col2.

I did not quite understand how implicit re-sorting would help remove all but first & last row for each unique value in Column1.

Note, the values in Column2 for each unique value in Column1 will always be in ascending order.

GOAL: To keep the MIN & MAX value form col2 in case there are intermediate values.

[Walter Roberson]

min() and max() is a different and easier calculation than removing values positionally. If you want min() and max() you can do your calculation using accumarray()