## creating matrix w.r.t. rows number of another matrix

### sermet (view profile)

on 6 Sep 2017
Latest activity Commented on by Stephen Cobeldick

on 7 Sep 2017

### James Tursa (view profile)

data=[1;2;8;9;11;12;19;22;24;26;30];
col_row=[1 2;1 4;1 8;1 11;2 4;2 8;2 11;3 10];
rows number in col_row are always ascending order. I need to store result_i matrix w.r.t. data and rows value of col_row matrix as follows.
result_1=[data(2);data(4);data(8);data(11)];
result_2=[data(4);data(8);data(11)];
result_3=data(10);
rows number in col_row always starts with 1. I need to extract second column values of each repeated numbers (1-2) and single number in col_row matrix and create result_i matrix.
Above data is example, actual data matrix is bigger and the maximum row value of col_row matrix is variable. How can I write a code to store result_i matrix?

James Tursa

### James Tursa (view profile)

on 6 Sep 2017
No, you don't want to do this. Creating variables with names like result_1, result_2, result_3 etc is a bad way of programming. It is hard to write downstream code generically to use these variables. The resulting code will be hard to read and hard to debug. Much better would be to use something like cell arrays, e.g. result{1}, result{2}, result{3} etc. These will be easy to use downstream in your code in loops etc.
Stephen Cobeldick

### Stephen Cobeldick (view profile)

on 7 Sep 2017
"I need to ... and create result_i matrix"
Do not do this. It will make your code complex, slow, buggy, hard to debug, and obfuscated. Read this to know why creating or accessing variable names dynamically is a bad idea:
Indexing is simple, efficient, neat, easy to debug, and easy to understand: you should use indexing with some kind of array: a matrix, or an ND array, or a cell array, or a structure, or a table, or....

### James Tursa (view profile)

on 6 Sep 2017
Edited by James Tursa

### James Tursa (view profile)

on 6 Sep 2017

E.g., a method using a cell array
u = unique(col_row(:,1));
n = numel(u);
result = cell(max(u),1);
for k=1:n
result{u(k)} = data(col_row(col_row(:,1)==u(k),2));
end
This assumes that all of the "row" numbers are positive integers. If they are not, then one could do this instead
u = unique(col_row(:,1));
n = numel(u);
result = cell(n,1);
for k=1:n
result{k} = data(col_row(col_row(:,1)==u(k),2));
end