bucketing values

Hi

I have a vector of values from 0 to 1 representing probabilities

A = [0.8756 0.1185 0.0059]

How can i generate a random number using rand() and map it to a value in A without looping. I can;t see how i can do this with vectorisation or with the built in functions.

My aim is generate a random numbers so 87% of the time I choose A(1), 11% A(2), etc...

Thanks

Answer 1

[Teja Muppirala]

If you have the Statistics Toolbox installed, there is the built-in RANDSAMPLE command to do this.

A = [0.8756 0.1185 0.0059];
X = randsample(numel(A),5000,true, A);  % X gets 5000 samples

tabulate(X)

Answer 2

[Daniel Shub]

Assuming your array A is short ...

A = [0.8756 0.1185 0.0059]
x = rand(1e3, 1);
y(x < A(1)) = 'a';
y(x >= A(1) & x < A(2)) = 'b';
y(x >= A(2)) = 'c';

[Fangjun Jiang]

I think all the answers provided are along the same line. In your case, the A needs to be [0.87,0.98,1].

[Daniel Shub]

You are correct. I was sloppy.

Answer 3

[Sean de Wolski]

Like this?

A = [0.8756 0.1185 0.0059]; %percentiles (must sum to 1
R = rand(1,1000); %random data
B = [11 17 19]; %sample data to extract A% of the time
[junk,Bin] = histc(R,[-inf cumsum(A)]);  %find the bin
C = B(Bin) %extract

Answer 4

[Fangjun Jiang]

Something like this:

A = [0.8756 0.1185 0.0059];
RandNum=rand(1000,1);
Index=2*ones(size(RandNum));
Index(RandNum<A(1))=1;
Index(RandNum>1-A(3))=3;

sum(Index==1)
sum(Index==2)
sum(Index==3)

ans =
   892
ans =
   102
ans =
     6

close enough?

Answer 5

[Greg]

Thanks for all the answser

I'm not sure which idea scales better as in general A[] will be of different sizes so I can't explicitly code for 3 items.

I'm fairly new to Matlab so any further pointers is much appreciated.

Thanks

[Sean de Wolski]

Randsample and Histc are both automated for large/varying A.

Answer 6

[Greg]

Thanks Teja

I have the Stats Toolbox so this looks to be exactly what i'm looking for. I will play around with this some more but so far so good

Cheers Greg