Matrix Sortrows by one of it Values

Hello Everyone

sorry for this simple question, But i Need answer ASAP,and i'm reall stuck in it

sorry for bad English also i just realized the whole question is wrong:| , but still problem Exist if You Still want to help :

i have a matrix it has about 150 rows in it , also i input a number to a variable,

i want search all of my matrix rows and find closest value to my variable value number i gave ,

i already wrote this part , problem is that first time it finds closest value in my matrix , i want to find second and third value close to my input value,

Thanks , for Your Reply

M.Dadmand

Answer 1

[Image Analyst]

Try this:

% Sample data
A =[...
9 0.28
3 0.50
1 0.30
4 0.35
2 0.65
6 0.25]

% Sort rows by how close the second column value is to 0.32.
[~, sortedIndexes] = sortrows(abs(A-0.32), 2);
% Create an output array that is A shuffled by the sort index.
out = A(sortedIndexes, :)

Use [sortedA, sortedIndexes] if your version of MATLAB is so old that it doesn't support the ~ syntax.

If you want to extract just the top 3 rows, use this instead:

out = A(sortedIndexes(1:3), :)

A =
   9                      0.28
   3                       0.5
   1                       0.3
   4                      0.35
   2                      0.65
   6                      0.25
out =
   1                       0.3
   4                      0.35
   9                      0.28

[Mohammad]

i think this is the answer, i still need to test it , my Data set is big, just have a question when Indexes get higher it column value gets more far from our input Var to compare and search, Yes?

i'll mark it answered as soon as i'll get sure it 100% works , Thank You so much. <3

[Image Analyst]

It should work no matter what value you're comparing against.

Answer 2

[Cedric]

You should play a bit with sortrows, just to have a better feeling of how it works. For example

 >> A = [2 7 12; 1 3 10; 2 6 13; 1 2 11]
 A =
     2     7    12
     1     3    10
     2     6    13
     1     2    11

Now I would like to sort this matrix first in ascending order based on column 1, and then (when multiple elements of the first column are identical) in descending order base on column 2:

 >> sortrows(A, [1, -2])
 ans =
     1     3    10
     1     2    11
     2     7    12
     2     6    13

The priority between columns for the sort is defined by elements of the 2nd argument, and the ascending or descending behavior is defined the sign of these elements.

Now your question doesn't seem to make sense, because A(:,7) is a column vector, which means that it has only one column, so it is unclear why you want to sort it according to the value in A(100,7), which should be 1 or -1 for it to work dimension-wise, and which would be sorted with the rest of column 7.

[Cedric]

Here is a bit more information that, I think, will be useful to you after reading your reply to Image Analyst. Say you have a vector that you want to sort according to how close values of its elements are from a given value; you could proceed as follows:

 >> a = [1, 9, 4, 2, 13] ;               % Vector.
 >> v = 5 ;                              % Given value.
 >> [~,ix] = sort(abs(a-v)) ;
 >> a(ix)
 ans =
     4     2     1     9    13

Here you see that a(ix) gives you a list of closest values from closest to farthest. To understand this, look for abs() and sort() in the help. For the latter, you will see that it outputs the index of sorted elements as a 2nd output argument. So what we do here, is that we sort absolute values of differences, and then we index the vector with the index returned by sort(). Most important, decompose the code above to understand what it does, i.e. evaluate the following and see what each expression outputs:

 >> abs(a-v)
 >> sort(abs(a-v))
 >> [s,ix] = sort(abs(a-v))
 >> [~,ix] = sort(abs(a-v))

Answer 3

[Image Analyst]

I have no idea what "sort it rows by the value is in A[100,7]" means - perhaps get another English speaker to paraphrase that. The help for sortrows() says this:

B = sortrows(A,column) sorts the matrix based on the columns specified in the vector column. If an element of column is positive, the MATLAB software sorts the corresponding column of matrix A in ascending order; if an element of column is negative, MATLAB sorts the corresponding column in descending order. For example, sortrows(A,[2 -3]) sorts the rows of A first in ascending order for the second column, and then by descending order for the third column.

Is there somethi9ng about that which you do not understand?

[Mohammad]

sorry for bad English also i just realized the whole question is wrong:| , but still problem Exist if You Still want to help :

i have a matrix it has about 150 rows in it , also i input a number to a variable,

i want search all of my matrix rows and find closest value to my variable value number i gave ,

i already wrote this part , problem is that first time it finds closest value in my matrix , i want to find second and third value close to my input value,

Thanks , for Your Reply

[Image Analyst]

See my new answer below.

Answer 4

[Azzi Abdelmalek]

A=randi(10,5,4)
value=4
[n,m]=size(A)
b=abs(A-value)
[jj,jj]=sort(b,2)
ii=repmat((1:n)',1,m)
idx=sub2ind(size(A),ii,jj)
out=A(idx)

[Mohammad]

thanks Mr.Abdelmalek You answered my another question before :D

I don't compeletly undrestand the code

here is another example i want to do if i have this :

A =

9 0.28
3 0.50
1 0.30
4 0.35
2 0.65
6 0.25]

and i input 3 values near 0.32 it should give me

A =[9,0.28;1,0.30;4,0.35]

[Azzi Abdelmalek]

Why the size of A is changing?

[Mohammad]

we can use another Variable

B = [9,0.28;1,0.30;4,0.35]

the whole project is simulation K-nearest neighbour algorithm so for example we say find 4 nearest column values to the value we get from input and it search second column of A and find first 3 nearest value .

[Image Analyst]

Why does the row with 9 come first? It is farther away than the row with 1 in the first column.

[Mohammad]

its not index its first row with some value in it. it actually doesnt matter whats in it , i wrote it because it is important when we find K nearest Values we need to keep and know our indexes,

[Image Analyst]

Uh, okay... Azzi - did you understand that? Because I don't. So neither Azzi's nor my script does what you want?

[Azzi Abdelmalek]

No, I did'nt