## How to count unique value in array?

### tedy (view profile)

on 6 Apr 2013

Hello, i have an example array a= [2,2,2,1,1,3,2,1,3,3,2] i do this code

```a= [2,2,2,1,1,3,2,1,3,3,2]
unique(a)
histc(a, unique(a))
```

but, it only result 1,2,3 and 3,5,3 as its frequency.

How to take all the value, but no one duplicate each others. it's like 2,1,3,2,1,3,2 and also with its frequency 3,2,1,1,1,2,1. Thanks in advance

## Products

No products are associated with this question.

### Cedric Wannaz (view profile)

on 6 Apr 2013
Edited by Cedric Wannaz

### Cedric Wannaz (view profile)

on 6 Apr 2013

You could build something around the following example:

``` >> b = a([diff(a)~=0, false])
b =
2     1     3     2     1     3```

let me know if you need help to get counts. Hint: try using DIFF as well as FIND.

tedy

### tedy (view profile)

on 6 Apr 2013

Thank you cedric. How about the frequency? i code with histc(b, unique(b)), but it result 2,3,2, not 3,2,1,1,1,2,1

Cedric Wannaz

### Cedric Wannaz (view profile)

on 6 Apr 2013

Look at the following:

``` >> diff(a)
ans =
0     0    -1     0     2    -1    -1     2     0    -1```

as you can see, places where diff is not null are places where there is a transition in the value. These locations can be found using FIND:

``` >> find(diff(a))
ans =
3     5     6     7     8    10```

Now you can probably imagine a solution based on DIFF again..

``` >> c = diff([0, find(diff(a)), numel(a)])
c =
3     2     1     1     1     2     1```

EDIT: I see that Roger posted a solution while I was writing my comment. His solution for getting counts is based on the same principle, but it is better than mine in the sense that it involves only two additional logicals, whereas mine involves a call the NUMEL.

### Roger Stafford (view profile)

on 6 Apr 2013
` f = diff(find([true,diff(a)~=0,true]));`

tedy

### tedy (view profile)

on 6 Apr 2013

Thank you so much Roger!

MATLAB and Simulink resources for Arduino, LEGO, and Raspberry Pi test