Delete row if element is out of specified range

I have the following:

min = 67;
max = 69;

nmat =

           0    0.9000    1.0000   64.0000   82.0000         0    0.5510
      1.0000    0.9000    1.0000   71.0000   89.0000    0.6122    0.5510
      2.0000    0.4500    1.0000   71.0000   82.0000    1.2245    0.2755
      2.5000    0.4500    1.0000   69.0000   70.0000    1.5306    0.2755
      3.0000    0.4528    1.0000   67.0000   72.0000    1.8367    0.2772
      3.5000    0.4528    1.0000   66.0000   72.0000    2.1429    0.2772
      4.0000    0.9000    1.0000   64.0000   70.0000    2.4490    0.5510
      5.0000    0.9000    1.0000   66.0000   79.0000    3.0612    0.5510
      6.0000    0.9000    1.0000   67.0000   85.0000    3.6735    0.5510
      7.0000    1.7500    1.0000   66.0000   72.0000    4.2857    1.0714
      9.0000    0.4528    1.0000   64.0000   74.0000    5.5102    0.2772
      9.5000    0.4528    1.0000   67.0000   81.0000    5.8163    0.2772
     10.0000    0.9000    1.0000   71.0000   83.0000    6.1224    0.5510
     11.0000    0.4528    1.0000   71.0000   78.0000    6.7347    0.2772
     11.5000    0.4528    1.0000   69.0000   73.0000    7.0408    0.2772
     12.0000    0.4528    1.0000   67.0000   71.0000    7.3469    0.2772
     12.5000    0.4528    1.0000   66.0000   69.0000    7.6531    0.2772
     13.0000    0.4528    1.0000   67.0000   83.0000    7.9592    0.2772
     13.5000    0.4528    1.0000   66.0000   72.0000    8.2653    0.2772
     14.0000    0.4528    1.0000   64.0000   74.0000    8.5714    0.2772
     14.5000    0.4528    1.0000   63.0000   70.0000    8.8776    0.2772
     15.0000    0.9000    1.0000   64.0000   82.0000    9.1837    0.5510
     16.0000    1.8028    1.0000   64.0000   75.0000    9.7959    1.1037

and I would like to delete any row in which the value for (i,4) is out of the particular range:

min < nmat(i,4) < max

I am aware that Matlab does not like the syntax a < b < c, but I have tried several variations of code and cannot understand where my error is.

min = 67;
max = 69;

for i=1:length(nmat)
if (nmat(i,4) < min) || (nmat(i,4) > max)
nmat(i, :) = [];
end
end

Logical operations aside, I cannot even get the loop to properly perform the more simple task of deleting rows with an (i,4) value below the min:

    min = 67;
    max = 69;

    for i=1:length(nmat)
    if nmat(i,4) < min
    nmat(i, :) = [];
    end
    end

Any help is much appreciated.

Answer 1

[Jan]

toDelete          = ((min < nmat(:,4)) | (nmat(:,4) < max));
nmat(toDelete, :) = [];

Your loop must fail, because it checks all rows of teh matrix, but this matrix shrinks during the processing. The limits of the loop are determined once only at the beginning.

Alternatively you could run the loop from size(nmat, 1) backward to 1. But the iterative changing of the array size wastes resources, so better avoid this.

[JZ]

Jan, why am I getting the following error with your code?

Subscript indices must either be real positive integers or logicals.

Answer 2

[Guillaume]

First a comment on your code:

for i = 1:length(nmat)

In my opinion, length should be removed from matlab. There are clearer alternatives and most people who use it don't understand how it works.

With a 2d matrix, length is the size of the largest dimension. So if the matrix has more rows than columns, it's the number of rows and if the matrix has more columns than rows, it's the number of columns.

In your particular case, length might start as the number of columns, but once you've deleted enough will switch the number of columns!

It's better to always be explicit. If you want the number of elements in a vector use numel, if you want the number of rows in a matrix use size(m, 1), and size(m, 2) to get the number of columns.

Second comment: Do not use min and max as variable names as you won't be able to use the min and max function of matlab.

Now, to answer your question, the simplest way to delete your rows is with:

thresholdmin = 67;
thresholdmax = 69;
nmat(nmat(:, 4) < thresholdmin | nmat(:, 4) > thresholdmax, :) = []

[JZ]

Thanks for the solution and explanation Guillaume. You answer works properly for my data but I am having trouble expanding the concept.

Lets say I want to remove rows with values in the slightly more complex range:

(less than 1 OR greater than 3 in column 1) AND (less than 70 in column 4)

What is incorrect about the following code:

col2min = 1;
col2max = 3;
col4max = 70;

nmat(nmat(:, 4) < col4min  &  nmat(:, 2) < col2min | nmat(:, 2) > col2max, :) = [];

I imagine it is some subtlety in the syntax, but I cannot catch it. Along those lines, I am not sure I fully understand why in your original code the specification

, :

is needed only after the second conditional

( a < tmin | b > tmax , : )

Would the syntax need to be different if we were implementing the & operator? For ex:

( a < tmin, : & b > tmax , : )

Thanks again

[Guillaume]

First, your expression does not work for two reasons, a) you've got the wrong column (2 instead of 1) and b) you've fallen foul of operator precedence. In matlab AND always takes precedence over OR, so your expression says:

(less than 70 in column 4 AND less than 1 in column 2) OR (greater than 3 in column 2)

The answer is to use parentheses just as you'd written in your statement, so:

nmat(nmat(:, 2) < col4min & (nmat(:, 1) < col2min | nmat(:, 1) > col2max), :) = []

The way this works, nmat(:, x) is a column vector with as many rows as nmat. Compared to a scalar you get a column vector of logicals. This is then combined with other column vectors of logicals with & or |, so in the end you have a column vector to select with row you want to pick / discard. To delete a whole row, you need to select all the columns, hence why you have : for the columns. In short, you're simply using

nmat(rowidx, colidx) = []

where rowidx is a logical vector to select which rows you want, and colidx is : to select all columns.

[JZ]

Thanks for all the explanations - It seems I have it working now.

Answer 3

[Stalin Samuel]

min = 67;
max = 69;
nmat = randi([1 500],100,100);%for example
[row,col] = find(nmat>min&nmat<max)
nmat(unique(row),:)=[]