# How can I convert 2 column matrix to a cell array?

2 views (last 30 days)
Jay Vaidya on 30 Dec 2019
Edited: Jay Vaidya on 1 Jan 2020
Is there a convenient way to convert a 2 column matrix into a cell array (without using nested for loops if possible)?
The matrix looks like:
Can I make a matrix from the above matrix such that the rows and columns of the new matrix(or a cell array you can say) will be like you see below:
the empty cells can be NAN and the filled cells are the values of the 3rd column of the old matrix.

Walter Roberson on 30 Dec 2019
result = accumarray(matrix(:,[2 1]), matrix(:,3), [], @(v) {v}, {nan})
However I wonder if you would be better off using
result = sparse(matrix(:,2), matrix(:,1), matrix(:,3))
which would be a sparse numeric array instead of a cell array that takes up about 108 bytes per entry.
##### 3 CommentsShow 1 older commentHide 1 older comment
Andrei Bobrov on 30 Dec 2019
[i,g] = findgroups(matrix(:,2));
out = flip(accumarray([matrix(:,1),i],matrix(:,3)),1);
Jay Vaidya on 1 Jan 2020
Edited: Jay Vaidya on 1 Jan 2020
That is what I am confused as well. If the input matrix(:,1:2) is of the order of mx2. I want to make a graph from this matrix. Lets say the order of the graph (look below) is pxq. Then, the order of the the result matrix should be pxq as well. In other words, I want to transform,
matrix->graph->adjacency matrix-> heatmap. My ultimate goal is to plot a heatmap from the input matrix. Mr. Bobrov, I have made a different thread for this, it would be great if you can help. https://www.mathworks.com/matlabcentral/answers/498570-how-to-convert-the-graph-below-in-to-heatmap

Jay Vaidya on 31 Dec 2019
Hello Mr. Roberson and Mr. Bobrov,
I am getting an error while running that. I have attached the snapshots of the error that says that SUBS must contain positive integer subscripts(associated with accumarray command) and another one that has an index value problem (associated with the sparse command). For convenience, I have attached the matrix.xlsx file that has the 'matrix' that is being used in this problem.
Attachment: matrix.xlsx (see below)
Thanks.

Andrei Bobrov on 31 Dec 2019
T.Var3 = str2double(T.Var3);
T = T(any(T{:,1:2} ~= 0,2),:);
M = T{:,:};
[i,g] = findgroups(M(:,2));
out = flip(accumarray([M(:,1),i],M(:,3)),1);
##### 3 CommentsShow 1 older commentHide 1 older comment
Andrei Bobrov on 31 Dec 2019
T.Var3 = str2double(T.Var3);
T = T(any(T{:,1:2} ~= 0,2),:);
M = T{:,:};
i = findgroups(M(:,2));
out = flip(accumarray([M(:,1),i],M(:,3)),1);
Jay Vaidya on 31 Dec 2019
So, the out matrix is of the order of 2550x2550. Where the graph is a 51(rows)x51(column) graph. Please see the image below. I want an 'out' matrix which has dimensions of 51x51 and represents just weights. In other words, the 'out' matrix will exactly the same as the graph in figure b.

### Categories

Find more on Sparse Matrices in Help Center and File Exchange

R2019b

### Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!