How do I delete the rows in amtrix wherre one column is less than another.

I have a matrix

[8 29 46 75;

8 29 58 50

8 29 58 75

8 46 58 75

29 46 58 50

29 46 58 75

8 29 46 25

8 46 58 25]

I need to find the rows where the fourth column is the highest value. For example 29 46 58 50

should be deleted and

29 46 58 75 should remain.

I have already tried unique but maybe I am not using enough qualifiers.

Also

MaxTripUnique = TripUnique((max(TripUnique(:,4))),:) Attempted to access TripUnique(75,:); index out of bounds because size(TripUnique)=[9,4].

with no luck.

Any help is appreciated as all the ideas on the internet don't seem to work.

This is part of a program where the matrix is created in a loop and I need to take the final values and concatenate them into the matrix that will be used for a further calcualtion.

Answer 1

[Joseph Cheng]

can be simply performed by

testmat = [8 29 46 75;8 29 58 50;8 29 58 75;8 46 58 75;29 46 58 50;29 46 58 75;8 29 46 25;8 46 58 25];
[val ind]=max(testmat,[],2)
newmat = testmat(ind==4,:)

so by using the dimension parameter inside the max() function we can tell it to perform the max per row and tell us which index the max was detected. Then using the variable ind we can define which rows has the maximum value in the 4th column.

[Thomas Sisk]

I tried this answer and then realized how would I adjust it if I had

8 29 46 75
8 46 58 50
8 46 58 25

This method would take just the first row and I would still need the second row as it is the larger value of the two bottom rows.

Any ideas?

[Stephen23]

@Thomas Sisk: You are introducing a new requirement which seems to be related to pairs or rows, something that your original question does not explain. As amazing as it might seem, we can't read your mind: Joseph Cheng's answer resolves the original question that you have asked. If you have some different or new conditions that need to be fulfilled, then you need to explain them clearly, preferably with examples too.

And please take the time to format your code properly: it make sit much clearer for us to work with:

http://www.mathworks.com/matlabcentral/answers/13205-tutorial-how-to-format-your-question-with-markup

[Joseph Cheng]

Sorry, been too busy to check Answers. I do not understand the new requirement. Your initial request was clear but I do not understand the condition that you'd keep the second row? Just because it is larger of the two last rows is not enough as your initial condition wipes it out. (also in your question you have that row deleted as your example).

Answer 2

[Brendan Hamm]

To delete the rows of A where the fourth column is not the largest value in that row, we just need to perform a logical comparison. To do this we need a matrix containing the 4th column which has the same size as the first 3 columns.

cmpMat = repmat(A(:,4),1,3); % A 3 column vector with the last column repeated
isFour = A(:,1:3) > cmpMat; % Logical with 1 where value is larger than 4th column

Now we need to check if any of the values in a row are larger than 4th column and delete it.

rowIdx = any(isFour,2); % Logical identifying rows with a value larger than 4th col
A(rowIdx,:) = [];

Note: max is returning to you the maximum value and this is why this does not work as an index. The second output of max is the index, but this uses linear indexing if passed a matrix.

[Thomas Sisk]

TripUniqueTest =

     8    29    46    75
     8    29    58    50
     8    29    58    75
     8    46    58    75
    29    46    58    50
    29    46    58    75
     8    29    46    25
    29    46    58    25

>> cmpMat = repmat(TripUniqueTest(:,4),1,3)

cmpMat =

    75    75    75
    50    50    50
    75    75    75
    75    75    75
    50    50    50
    75    75    75
    25    25    25
    25    25    25

>> isFour = TripUniqueTest(:,1:3)>cmpMat

isFour =

     0     0     0
     0     0     1
     0     0     0
     0     0     0
     0     0     1
     0     0     0
     0     1     1
     1     1     1

>> rowIdx = any(isFour,2)

rowIdx =

     0
     1
     0
     0
     1
     0
     1
     1

>> TripUniqueTest(rowIdx) = []

TripUniqueTest =

Columns 1 through 11

     8     8     8    29    29    29    29    46    46    46    29

Columns 12 through 22

    46    46    58    58    58    58    58    46    58    75    50

Columns 23 through 28

    75    75    50    75    25    25

The above is what I got using this method any ideas whats wrong.

[Brendan Hamm]

Sorry, that last line should read:

TripUniqueTest(rowIdx,:) = [];