## how get graph?

### huda nawaf (view profile)

on 20 Apr 2012

hi if I have data show users nd the relationships with other users, are there any way by which can draw such graph?

### Image Analyst (view profile)

on 20 Apr 2012

I think you should know this, but I'll just throw out a wild guess: a scatter plot? If that's not what you want then you need to be more descriptive. These may help with that:

The "gray level co-occurrence matrix" (performed by the graycomatrix function in the Image Processing Toolbox) is one way of seeing relationships such as this where you want to see the relationship between every item and every other item. It's an image of "connection weights" though (like how many times those items occur together as a pair), rather than a graph-theory type of node diagram. It may or may not be helpful to you, since I tend to look at things through "imaging glasses." For example:

```% Create sample data.
m = uint8([1,100;1,3;1,10;2,100;3,2;3,4])
% Computer glcm matrix.
gcm = graycomatrix(m, 'numlevels', 255)
imshow(gcm, []);
axis on;
title('Gray Level Co-occurrence Matrix', 'Fontsize', 30);
xlabel('Column 1 Value', 'Fontsize', 30);
ylabel('Column 2 Value', 'Fontsize', 30);
% Enlarge figure to full screen.
set(gcf, 'units','normalized','outerposition',[0 0 1 1]);
```

glcm is like a scatterplot where the intensity of the dots depends on the frequency of occurrence of that pair.

### Walter Roberson (view profile)

on 20 Apr 2012

This sounds like a traditional kind of graph-theory plot. Unfortunately once you get to about 6 nodes, you cannot necessarily draw the result as a planar graph, so you are going to have to be more specific about how you want the graphs drawn.

Oleg Komarov

### Oleg Komarov (view profile)

on 22 Apr 2012

@Walter: your help is much needed in places where (at least) effort is.

Geoff

### Geoff (view profile)

on 23 Apr 2012

Sounds like the required solution to this problem is to write a program which generates a GraphViz script, then install GraphViz and run the script.

huda nawaf

on 24 Apr 2012