Is there a more elegant replacement for this MATLAB loop?

Question

I want to get better at vectorizing my loops in MATLAB. At the moment, I'm trying to count the occurrences of values in a list of ints. My code is similar to this:

list = [1 2 2 3 1 3 2 2 2 1 5];
occurrence_list = zeros(1,max(list));

for x=list
    occurrence_list(x) = occurrence_list(x) + 1;
end

Is there a simple vectorized replacement for that for loop? (Or is there a built in MATLAB function that I'm missing?) I'm doing this on pretty small data sets, so time isn't an issue. I just want to improve my MATLAB coding style. Thanks for any suggestions!

Answer 1

You can use the hist function for that. Specify an output, and force the bins to be integers over the range of your array.

list = [1 2 2 3 1 3 2 2 2 1 5]; 
bins = min(list):max(list);
counts = hist(list,bins);

Answer 2

So, this is basically a histogram. Off my memory - look for the HIST function.

Answer 3

In addition to the HIST/HISTC functions, you can use the ACCUMARRAY to count occurrence (as well as a number of other aggregation operations)

counts = accumarray(list(:), 1)
%# same as: accumarray(list(:), ones(size(list(:))), [], @sum)

Another way is to use TABULATE from the Statistics Toolbox (returns value,count,frequency):

t = tabulate(list)
t =
            1            3       27.273
            2            5       45.455
            3            2       18.182
            4            0            0
            5            1       9.0909

---

Note that in cases where the values don't start at 1m or in case there's large gaps between the min and the max, you will get a lot of zeros in-between the counts. Instead use:

list = [3 11 12 12 13 11 13 12 12 12 11 15];
v = unique(list);
table = [v ; histc(list,v)]'

table =
     3     1
    11     3
    12     5
    13     2
    15     1

representing the unique values and their counts (this will only list values with at least one occurrence)